EP0554035B1 - Solid state color video camera - Google Patents

Description

BACKGROUND OF THE INVENTIONField of the invention
• The present invention relates to a color video camera,particularly to the method of generating luminance signalsthereof.
Description of Related Art
• Fig.1 shows an arrangement of color filters of an imagesensor described, for example, on pages 76 through 82 of"National Technical Report", Vol. 31, No.1, Feb. 1985,composed by Matsushita Techno Research Co., Ltd. and publishedby Ohmsha Publishing Co., Ltd. In Fig.1, Mg representsa pixel having a magenta color filter, G represents apixel having a green color filter, Cy represents a pixelhaving a cyan color filter and Ye represents a pixel havinga yellow color filter. Fig.2 shows a part of a signal processingcircuit of a color video camera which employs animage sensor consisting of these color filters arrangedthereon. In Fig.2,numeral 1 indicates a lens,numeral 2indicates an image sensor,numeral 3 indicates a band-passfilter (BPF),numeral 4 indicates a detector,numeral 5indicates a one horizontal period delay circuit (1HDLY),numeral 6 indicates a switching circuit andnumeral 45indicates a low-pass filter (LPF).
• The operation will now be described below. In Fig.2,incident ray on thelens 1 forms an image on theimagesensor 2. In Fig.1, an output signal from line n of theimage sensor 2, consisting of a sequence of (Mg+Cy) and(G+Ye) being repeated, is denoted as Sn, and an outputsignal from line n+1 of theimage sensor 2, consisting of asequence of (Mg+Ye) and (G+Cy) being repeated, is denoted asSn+1. Then Sn and Sn+1 are represented by the followingequations.Sn = Yn+Cn · sin(ωt) + ...Sn+1 = Yn+1 + Cn+1 · sin(ω t) + ...Where ω is the carrier frequency of the color signal whichcorresponds to double the horizontal pixel width. Yn andYn+1 in equations (1) and (2) represent the luminance signalcomponents of line n and line n+1, Cn and Cn+1 represent thecolor difference signal components of line n and line n+1,respectively, and are given by the following equations.Yn = (Ye+G) + (Cy+Mg) = 2R+3G+2BYn+1 = (Ye+Mg) + (Cy+G) = 2R+3G+2BCn = (Cy+Mg) - (Ye+G) = 2B-GCn+1 = (Ye+Mg) - (Cy+G) = 2R - GThe luminance signal components Yn, Yn+1 are obtained bypassing the output of theimage sensor 2 through the low-pass filter 45. The color difference signal components Cn,Cn+1 are obtained by passing the output of theimage sensor2 through the band-pass filter 3 having a center frequencyω and adetector 4. The output of thedetector 4 gives 2R-Gand 2B-G appearing every two lines. Thesesignals 2R-G, 2B-Gwhich appear on every other line are synchronized by the onehorizontalperiod delay circuit 5 and theswitching circuit6.
• In the conventional color video camera as describedabove, an output signal of the image sensor is passedthrough a low-pass filter to remove the modulated componentsof the color signal and obtain a luminance signal, resultingin a problem of the harmonics of the luminance signals beingattenuated. Although aperture correction has been made byenhancing the rising edge and falling edge of the signal toimprove the resolution, it causes an impression of unnaturalenhancement.
• Fig.3 shows a block circuit diagram illustrating thecircuit of a color video camera employing a spatial offsetof 3-chip CCD color camera which is described, for example,on pages 1079 through 1085 of the "Journal of TelevisionEngineering Association", Nov. 1986. In Fig.3,numeral 51indicates a lens,numeral 52 indicates a refracting prismwhich decomposes incident ray into three colors of red,green and blue,numeral 53, 54, 55 indicate image sensors,numeral 56 indicates a red signal amplifier,numeral 57indicates a green signal amplifier,numeral 58 indicates ablue signal amplifier,numeral 68 indicates a low-pass filter,numeral 71 indicates an adder andnumeral 73 indicatesa demultiplexer. Fig.4 shows the constitution of output signalswith the conventional method of spatial offset of 3-chipCCD color camera. In Fig.4, G represents the signal ofa green pixel, and RB represents the composite signal of redand blue pixels. Letter p indicates the horizontal pixelwidth of the image sensor. Green image sensors and red, blueimage sensors are arranged in the horizontal direction atintervals of a half pixel width.
• The operation will now be described below. In Fig.3,incident ray on thelens 51 is decomposed into red, greenand blue by the refractingprism 52, with the light rays ofrespective colors forming images onimage sensors 53, 54,55. Each of theimage sensors 53, 54, 55 mixes signals ofthe upper and lower adjacent pixels to give one signaloutput. The output signals R, G, B of theimage sensors 53,54, 55 are amplified by thered signal amplifier 56, thegreen signal amplifier 57 and theblue signal amplifier 58,respectively, so that the ratio of the output signals thereofbecomes, in the case of NTSC system,
  R:G:B=0.30:0.59:0.11, to obtain R', G', B' signals. R' andB' are mixed in theadder 71 to obtain a signal RB which combines R' and B'. Thedemultiplexer 73 switches alternately between G' andRB to produce an output of luminance signal, which is passed through the low-passfilter 68 to obtain a luminance signal Y. Consequently, the luminancesignal Y is given by equation (7).Y = 0.30R + 0.59G + 0.11B= R' + G' + B'= G' + RB
• In the conventional method of spatial offset of 3-chip CCD color camera,the purpose is set at improving the resolution. Although there arises no problemin the case of such objects that have green signal and red-blue combined signal insimilar proportions, but vertical lines appear in the case of objects which havesignificantly different proportions. Vertical lines have been reduced by passingthe green signal and red-blue combined signal through a low-pass filter in theprior art, although it has a problem of causing attenuation of harmonics in theluminance signal. Thus resolution has been improved by enhancing the risingedge and falling edge of a signal for aperture correction, resulting in a problem ofunnatural enhancement.
• IEEE Transaction on Electron Devices, vol. 38, no. 5, May 1991, N.Ozawa and K. Takahashi "A Correlative Coefficient Multiplying (CCM) Methodfor Chrominance Moire Reduction in Single-Chip Color Video Cameras"discloses a method for reducing chrominance moire in a color video camera.One color signal is interpolated using a second color signal multiplied by a correlation coefficient related to the ratio of low-frequency components of thetwo color signals.
SUMMARY OF THE INVENTION
• It is desirable to provide a color video camera which is capable ofalleviating the attenuation of the harmonics of luminance signal.
• It is also desirable to provide a color video camera made of low costcircuits which is capable of producing high quality image with less attenuation ofthe harmonics of luminance signal.
• It is also desirable to provide a color video camera which is capable ofperforming aperture correction without unnatural enhancement by means ofsimple circuits at a low cost.
• The present invention provides a color video camera as set out inclaim 1.
• The invention also provides a color video camera as set out inclaim 17.
• The invention also provides a color video camera as set out inclaim 33.
• The invention also provides a method of deriving a signal representingluminance of an image as set out inclaim 34.
• The invention also provides a color video camera as set out inclaim 35.
• A low-pass filter may either be a one-dimensional low-pass filterconsisting of a plurality of bit-shift circuits and an adder, or a two-dimensionallow-pass filter consisting of a plurality of bit-shift circuits and a plurality ofadders.
• Use of a lookup table for division, a lookup table for logarithm, a lookuptable for power or the like simplifies the constitution of the signal processingcircuit.
• The above and further features of the invention will more fully beapparent from the following detailed description with accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
• Fig. 1 shows an example of the arrangement of color filters of an imagesensor used in a color video camera.
• Fig. 2 shows a part of signal processing circuit of the conventional colorvideo camera.
• Fig. 3 shows a block circuit diagram illustrating a color video camera ofthe conventional spatial offset of 3-chip CCD color camera system.
• Fig. 4 shows the constitution of the output signals of the conventionalspatial offset of 3-chip CCD color camera system.
• Fig. 5 shows a block circuit diagram illustrating a color video camera ofan embodiment of the present invention.
• Fig. 6 shows the signals written in the field memory.
• Fig. 7 shows the signals written in the two-dimensional memory.
• Fig. 8 shows the signals written in the two-dimensional memory.
• Fig. 9 shows the signals written in the two-dimensional memory.
• Fig. 10 shows the signals written in the two-dimensional memory.
• Fig. 11 shows the output signals of the two-dimensional low-pass filter.
• Fig. 12 shows the output signals of the two-dimensional low-pass filter.
• Fig. 13 shows the output signals of the two-dimensional low-pass filter.
• Fig. 14 shows the output signals of the two-dimensional low-pass filter.
• Fig. 15 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the present invention.
• Fig. 16 shows a block diagram illustrating the constitution of thearithmetic logic unit.
• Fig. 17 shows the signals written in the one-dimensional memory.
• Fig. 18 shows the signals written in the one-dimensional memory.
• Fig. 19 shows the output signals of the one-dimensional low-pass filter.
• Fig. 20 shows the output signals of the one-dimensional low-pass filter.
• Fig. 21 shows the output signals of the one-dimensional low-pass filter.
• Fig. 22 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the present invention.
• Fig. 23 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the present invention.
• Fig. 24 shows the constitution of the lookup table for division.
• Fig. 25 shows the constitution of the lookup table for division.
• Fig. 26 shows the constitution of the one-dimensional low-pass filter.
• Fig. 27 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the present invention.
• Fig. 28 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the present invention.
• Fig. 29 shows a block circuit diagram illustrating the constitution of thearithmetic logic unit.
• Fig. 30 shows the output signals of the two-dimensional low-pass filter.
• Fig. 31 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 32 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 33 shows the constitution of the two-dimensional low-pass filter.
• Fig. 34 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 35 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 36 shows a block circuit diagram illustrating the constitution of thearithmetic logic unit.
• Fig. 37 shows the signals written in the one-dimensional memory.
• Fig. 38 shows the signals written in the one-dimensional memory.
• Fig. 39 shows the signals written in the one-dimensional memory.
• Fig. 40 shows the output signals of the one-dimensional low-pass filter.
• Fig. 41 shows the output signals of the one-dimensional low-pass filter.
• Fig. 42 shows the output signals of the one-dimensional low-pass filter.
• Fig. 43 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 44 shows a block circuit diagram illustrating the constitution of thearithmetic logic unit.
• Fig. 45 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 46 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 47 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 48 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 49 shows the signals written in the two-dimensional memory.
• Fig. 50 shows the signals written in the two-dimensional memory.
• Fig. 51 shows the signals written in the two-dimensional memory.
• Fig. 52 shows the output signals of the two-dimensional low-pass filter.
• Fig. 53 shows the output signals of the two-dimensional low-pass filter.
• Fig. 54 shows the output signals of the two-dimensional low-pass filter.
• Fig. 55 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 56 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 57 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 58 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 59 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 60 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 61 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 62 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 63 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 64 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 65 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 66 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 67 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 68 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 69 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 70 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 71 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 72 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 73 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 74 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 75 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
• Fig. 76 shows a block circuit diagram illustrating a color video camera ofanother embodiment of the invention.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
• The invention will be described in detail below with reference to thedrawings which illustrate the preferred embodiments.
Embodiment 1
• Fig. 5 shows the block circuit diagram illustrating the color video camerainembodiment 1. In Fig. 5,numeral 1 indicates a lens, numeral 2 indicates an image sensor, numeral 3 indicates a band-pass filter (BPF),numeral 4 indicates adetector, numeral 5 indicates a one horizontal period delay circuit (1HDLY),numeral 6 indicates a switching circuit, numeral 7 indicates an A/D converter,numeral 8 indicates a field memory,numeral 9 indicates a multiplexer,numerals10, 11, 12, 13 indicate two-dimensional memories,numerals 14, 15, 16, 17indicate two-dimensional low-pass filters (two-dimensional LPF), numeral 42indicates an arithmetic logic unit and numeral 43 indicates a matrix circuit.
• The operation will now be described below. Incident ray on thelens 1 forms an image on theimage sensor 2 havingphotoelectric transducers which have different spectral-responsecharacteristics and are arranged on a two-dimensionalplane. The output of theimage sensor 2 isconverted from analog to digital signal by the A/D converter7 and is supplied to thefield memory 8. Theimage sensor 2shown in Fig.5 mixes the signals of two upper and loweradjacent pixels to give an output. Mixed signals are writtenin thefield memory 8 as one signal. Fig.6 partially illustratesthe configuration of writing the signals from theimage sensor 2 in thefield memory 8. In Fig.6, MC representsa signal which combines the magenta pixel signal and.cyan pixel signal, GY represents a signal which combines thegreen pixel signal and yellow pixel signal, MY represents asignal which combines the magenta pixel signal and yellowpixel signal and GC represents a signal which combines thegreen pixel signal and cyan pixel signal. Themultiplexer 9supplies the MC, GY, MY, GC signals separately to the two-dimensionalmemories 10, 11, 12, 13, respectively. Fig.7,Fig.8, Fig.9, Fig.10 show the configuration of signalswritten in the two-dimensional memories 10, 11, 12, 13. Thesignals shown in these drawings are smoothed by the two-.dimensional low-pass filters 14, 15, 16, 17. Fig.11, Fig.12,Fig.13, Fig.14 show the outputs of the two-dimensional low-passfilters 14, 15, 16, 17. LPF in the drawing represents a low-pass filter output.
• Arithmetic logical operation in thearithmetic logicunit 42 will be described below. In Fig.6, the value of anoutput signal, when color filter of MC, GY, MY are assumedto be located at the position of GC of row s and column tfor example, is calculated by equations (8) through (10).MC(s,t) = GC(s,t) × (MCLPF(s,t)/GCLPF(s,t))GY(s,t) = GC(s,t) × (GYLPF(s,t)/GCLPF(s,t))MY(s,t) = GC(s,t) × (MYLPF(s,t)/GCLPF(s,t))It is not necessary to calculate the value of GC(s,t) becauseequation (11) holds.GC(s,t) = GC(s,t)
• When it is assumed that a color filter of kind K (K iseither MC, GY, MY or GC) is located at the position (s, t)of a color filter of kind J (J is either MC, GY, MY or GC),color signal K (s,t) is calculated by equation (12). (s,t)in the case of this embodiment indicates the coordinates ofthefield memory 8 shown in Fig.6.K(s,t) = J(s,t) × (KLPF(s,t)/JLPF(s,t))
• Matrix computation in thematrix circuit 43 is describedbelow. Luminance signal component Y at the positionof GC at row s, column t in Fig.6 is calculated by equation(13).Y = (MC(s,t)+ GY(s,t)+ MY(s,t)+GC(s,t))/4
• The principle of calculating Y signal in this embodiment will be described below. If modulation component of acolor signal is removed by passing the output signal of theimage sensor through the low-pass filter as in the priorart, harmonics of the luminance signal will be attenuated.The method of this embodiment assumes that color change isnot significant in a localized region. This implies thatratios between signals of different colors (MC, GY, MY, GC)are approximately equal to each other in the localizedregion. Ratios between signals of different colors MC, GY,MY and GC in a localized region are given by the ratiosbetween two-dimensional low-pass filter outputs of MC, GY,MY and GC. When a color filter of MC is located at theposition GC of row s, column t as shown by equation (8), forexample, the ratio is obtained by multiplying GC(s,t) by theratio of MC to GC at the localized region (ratio of two-dimensionallow-pass filter output of MC signal to two-dimensionallow-pass filter output of GC signal). Equations(9), (10), (11) are also obtained similarly. By calculatingY signal by equation (13) using MC, GY, MY, GC obtained fromequations (8), (9), (10), (11), modulation component of thecolor signal can be removed without using the low-pass filters.Because modulation component of the color signal isremoved without using the low-pass filters, it is madepossible to alleviate the attenuation of the harmonics ofthe luminance signal.
Embodiment 2
• Fig.15 showsembodiment 2 of the invention where portionsdenoted with the same numerals as those in Fig.5represent the same or corresponding portions. In Fig.15,numerals 18, 19 represent one-dimensional memories,numerals20, 21, 22 represent one-dimensional low-pass filters (LPF)and numeral 23 represents an arithmetic logic unit. Fig.16shows the internal constitution of thearithmetic logic unit23, which has ademultiplexer 24, adivider 25 and amultiplier26.
• The operation will now be described below. Similarly tothe case ofembodiment 1 as shown in Fig.6, signal from theimage sensor 2 is written in thefield memory 8. Themultiplexer9 supplies MC, GY signals and MY, GC signals alternatelyto thememories 18, 19 via line n andline n+1.Fig.17 and Fig.18 show the configuration of signals writtenin thememories 18, 19. The signals shown in these drawingsare smoothed by the low-pass filters 20, 21. Fig.19, Fig.20show the outputs of the low-pass filters 20, 21. Output oftheimage sensor 2 is converted from analog to digitalsignal which is passed through the low-pass filter 22, toobtain low-pass filter output of Y signal. Fig.21 shows theoutput of the low-pass filter 22, where LPF in the drawingis a symbol representing a low-pass filter output.
• Arithmetic logical operation of thearithmetic logic unit 23 will be described below. The low-pass filter outputof Y signal is fed to thedivider 25 as the dividend. Thedemultiplexer 24 switches, in the case of line n, to thelow-pass filter output of MC if the pixel is MC pixel or tothe low-pass filter output of GY if the pixel is GY pixeland, in the case ofline n+1, switches to the low-passfilter output of MY if the pixel is MY pixel or to the low-passfilter output of GC if the pixel is GC pixel, with theoutput thereof being fed to thedivider 25 as the divisor.The output of thedivider 25 is supplied to themultiplier26 as the input. The output signal of the pixel is suppliedto themultiplier 26. As a result, the output of themultiplier 26 is obtained as Y signal at the pixel ofinterest.
• In Fig.17, luminance signal component Y (t) at theposition of color filter GC of column t, for example, iscalculated by equation (14) below.Y(t) = GC(t) × (YLPF(t)/GCLPF(t))
• Luminance signal component Y (t) at the position t ofcolor filter of kind K (K is either MC, GY, MY or GC) iscalculated by equation (15) below. In this embodiment,letter t represents the coordinate ofmemory 18 shown inFig.17 in case the pixel is color filter of GC, GY, orrepresents the coordinate ofmemory 19 shown in Fig.18 incase the pixel is MY, MC.Y(t) = K(t) × (YLPF(t)/KLPF(t))
• Principle of calculating Y signal inembodiment 2 willbe described below. This method assumes that color change inthe localized region is not significant. This implies thatthe ratios of signals of respective colors (MC, GY, MY, GC)to Y signal are approximately equal in a localized region.Ratios of signals of different colors MC, GY, MY and GC to Ysignal in a localized region are given by the ratios of low-passfilter outputs of MC, GY, MY and GC to the low-passfilter output of Y signal.
• For example, luminance signal component Y(t) at theposition of GC of column t as shown by equation (14) isobtained by multiplying GC(t) by the ratio of Y to GC (ratioof the low-pass filter output of Y signal to the low-passfilter output of GC signal) in the localized region.
• While harmonics of luminance signal is attenuated inthe conventional method as described forembodiment 1,modulated components of color signal can be removed withoutattenuating the harmonics of the luminance signal inembodiment2 as inembodiment 1.
Embodiment 3
• Although constitution of the color video camera inembodiment 3 is the same as that of embodiment 2 (Fig.15),method of calculating Y signal in thearithmetic logic unit 23 is different.
• Arithmetic operation in thearithmetic logic unit 23will be described below. In Fig.17, if each signal is madeup of eight bits to represent the hue in 256 steps, forexample, andvalue 1 of LSB (Least Significant Bit) isemployed as a constant, luminance signal component Y(t) atthe position of color filter GC of column t is calculated byequation (16) as shown below.Y(t) = (GC(t)+1) × ((YLPF(t)+1)/(GCLPF(t)+1))- 1
• Luminance signal component Y (t) at the position t ofcolor filter of kind K (K is either MC, GY, MY or GC) isgiven by equation (17) below.Y(t) = = (K(t)+1) × ((YLPF(t)+1)/(KLPF(t)+1))- 1
• There may arise such a case as calculation of divisionbecomes impossible due to the divisor being zero, or a greatcalculation error is caused by the divisor being not equalto but near zero if calculating with a small number of bits.In such a case, calculation error can be decreased byemployingvalue 1 of LSB as inembodiment 3.
Embodiment 4
• Fig.22 shows a block circuit diagram illustrating acolor video camera ofembodiment 4. In Fig.22, portionsdenoted with the same numerals as those in Fig.15 represent the same portions.Numeral 27 represents a comparator.
• The operation will now be described below. An outputsignal of an appropriate pixel in the vicinity of the pixelof interest is read from thefield memory 8 and is fed tothecomparator 27. If the difference between the outputsignals of pixels of the same kind is beyond a particularthreshold, it is determined that the spatial frequency ofthe image is high and, if the difference is within thethreshold, it is determined that the spatial frequency ofthe image is low. In a portion where the spatial frequencyof the image is high, the luminance signal is calculatedsimilarly toembodiment 2 and, in a portion where the spatialfrequency of the image is low, the luminance signal iscalculated from the weighted averaging value of the pixelsof N kinds around the pixel of interest.
• Thearithmetic logic unit 23 operates as inembodiment2 in the portion of high spatial frequency and, in theportion of low spatial frequency, luminance signal componentY(t) is calculated by equation (18) at the position of GCcolor filter in column t in Fig.17 and at the position wherethere is no output signal of column t by equation (18).Y(t) = MY(t-1)/4+GC(t)/2+MY(t+1)/4
• Assuming the position of the pixel being t, the kind ofcolor filter of the pixel of interest being J (J is eitherMC, GY, MY or GC), and the kind of color filter of the right and left adjacent pixels of the pixel of interest being K (Kis either MC, GY, MY or GC), the luminance signal componentY(t) is calculated by equation (19).Y(t) = K(t-1)/4+J(t)/2+K(t+1)/4
• As described above, the edge-like effects due to theimage contrast becoming exaggerated producing emphasizedpatches of light and dark, in the low frequency componentsarising when the number of bits of calculated data is reduced,can be suppressed by employing different methods incalculating the luminance signals between portions of highspatial frequency and low spatial frequency.
Embodiment 5
• Although the constitution of the color video camera inembodiment 5 is the same as that of embodiment 4 (Fig.22),signal processing procedures in thearithmetic logic unit 23and in thecomparator 27 are different.
• Inembodiment 5, difference in the output signals betweenright and left adjacent pixels of the pixel of interestand the particular threshold are compared. If the differencebetween the output signals is greater than thethreshold, it is determined that the portion has a highspatial frequency and the operation ofaforementioned embodiment2 is carried out, and, if the difference between theoutput signals is less than the threshold, it is determined that the portion has a low spatial frequency, and the operationofaforementioned embodiment 4 is carried out.
• Arithmetic operation of thecomparator 27 will bedescribed below. Assuming that, for example, the colorfilter located at row s, column t is GC and the color filterslocated at right and left adjacent pixels of the pixelof interest are MY, the result of the calculation by equation(20) below and the threshold are compared to select oneof the two equations to generate the luminance signals.| MY(t-1) - MY(t+1) |
• Assuming that the position of the pixel of interest ist and the kind of color filters of the right and left adjacentpixels of the pixel of interest being K (K is eitherMC, GY, MY or GC), result of the calculation by equation(21) below and the threshold are compared to select one oftwo equations to generate the luminance signals.|K(t-1) - K(t+1) |
Embodiment 6
• Fig.23 shows a block circuit diagram of the color videocamera inembodiment 6. In Fig.23, numerals which are thesame as those in Fig.15 indicate the identical portions.Numeral 28 represents a lookup table for division.
• Since the equation of calculation described inembodiment2 includes a dividing operation, the lookup table fordivision 28 is used inembodiment 6. When 55/13 is to be calculated with 5-bit input to the lookup table fordivision28 assuming that 8-bit inputs of 55 (00110111) and 13(00001101) are given as shown in Fig.24, for instance, upperfive bits of each input are checked successively startingwith the most significant bit to search for a bit of avalue1 and, if found, it is made the new most significant bitwith the five bits including this and those that followhandled as the input. Namely, inputs are bit-shifted toobtain the most effective five bits such as 27 (11011) and13 (01101). Then output 2 (00010) which corresponds to theresult of division of the two inputs is obtained from thelookup table fordivision 28, and the number of digitsthereof is raised by the number of digits which were cut offfrom the dividend, thereby to obtain 4 as the result of thearithmetic operation.
• Assume m-bit inputs x and y and n-bit input to thelookup table for division 28 (m>n), and the most significantbit of the input is checked to see if it is 1. If it is 0,lower bits are checked successively stopping at the nth bit. ,If x has 1 in the ath bit (m ≧ a ≧ n} and y has 1 in the bthhit (m ≧ b ≧ n), these bits are regarded as the most significantbits thereby to obtain the upper n bits x' and y'.Namely, x' and y' are the most effective n bits with theless significant bits thereof being discarded as shown byequations (22) and (23).x' = x>>(a-n)y' = y>>(b-n)z' = x'/y'Where » represents a bit shift operation, with the digitsof x being reduced by a-n bits and the digits of y beingreduced by b-n bits. With the resultant x' and y' being fedas the inputs, an output z' is obtained from the lookuptable fordivision 28, and bit shift operation correspondingto the number of bits which were cut off from the dividendand divisor is applied to z'. Namely, operation of equation(25) is carried out to obtain the result z.z = z' >> (b-a)
• Let the output of the lookup table fordivision 28 be anumber of eight bits, for instance, then dividing operationof the upper seven bits of the two inputs require the lookuptable fordivision 28 to have a capacity of 131072 bits,while similar result can be obtained with the lookup tablefordivision 28 having a capacity of 8192 bits when mosteffective five bits of inputs are used in the dividingoperation. As a result, circuit composition ofembodiment 6is made simpler.
Embodiment 7
• Although the composition of the color video camera inembodiment 7 is the same as that of embodiment 6 (Fig.23),signal processing operations in thearithmetic logic unit 23 and the lookup table fordivision 28 are different.
• The operation will now be described below. Lookup tablefordivision 28 in embodiment 7 has five bits above decimalpoint and five bits below decimal point. In calculating thedivision of 55/13 with 5-bit inputs to the lookup table fordivision 28 when 8-bit inputs of 55 (00110111) and 13(00001101) are given as shown in Fig.25, for instance,inputs are bit-shifted to the most effective five bits suchas 27 (11011) and 13 (01101). Then output 66 (0001000010)which corresponds to the result of the division of the twoinputs is obtained from the lookup table fordivision 28,and its number of digits is raised by the number of digitswhich were cut off from the dividend. Thus 132 is obtainedas the result of the arithmetic operation. Then this resultof the division is used in the multiplication by the equationdescribed inembodiment 2, with the lower five bitsbeing cut off to discard the fractional part of the lookuptable fordivision 28. Assuming the value of the multiplierto be 11, 45 is obtained from the above calculation.
• Assume that m-bit inputs x and y are given, an n-bitinput is fed to the lookup table for division 28 (m>n) toobtain a 2n-bit output comprising an integral part of n bitsand a fractional part of n bits (m>n), x includes the first1 in the ath bit from the MSB (m ≧ a ≧ n), y includes thefirst 1 in the bth bit from the MSB (m ≧ b ≧ n), most effective n-bit parts of x and y are x' and y', respectively,which are fed to the lookup table fordivision 28 as inputsto obtain an output z', and bit shift is applied correspondingto the number of bits which were cut off the dividendand divisor, to obtain z as the result of calculation. Thenthe value of z is used in the multiplication described inembodiment 2 with the result being shifted down by n bits todiscard the fractional part of the lookup table fordivision28. Assuming the value of multiplier being p, the result ofcalculation q is given by equation (26) as shown below.q = (p× z) >>n
• As described above, precision of calculation can beincreased by obtaining an output with the number of bitstwice that of the input from the lookup table fordivision28 to obtain the fractional part as the output, then multiplyingwith this output.
Embodiment 8
• The composition of the color video camera inembodiment8 is the same as that of embodiment 2 (Fig.15). Inembodiment8, low-pass filter is used as a digital filter. Thelow-pass filter is made up of only bit shift circuits suchas, for example, the number of horizontal taps being set to5 withweightings 1/8, 1/4, 1/4, 1/4 and 1/8. Fig.26 showsthe composition of the one-dimensional low-pass filter. In Fig.26, 101 through 104 are one clock delay circuits(1CLKDLY), 105 and 109 are 3-bit shift circuits, 106 through108 are 2-bit shift circuits, and 110 is an adder.
• Assume that an image sensor output S(t+2) is fed to theone-dimensional low-pass filter of such a composition. The3-bit shift circuit 105 feeds S(t+2)/8 to theadder 110.Output S(t+1) of the image sensor delivered at the time oneclock earlier is fed via the one clock delay circuit 101 tothe 2-bit shift circuit 106 to obtain an output S(t+1)/4.Similarly, outputs S(t)/4, S(t-1)/4 and S(t-2)/8 are obtainedfrom the 2-bit shift circuits 107, 108 and the 3-bitshift circuit 109. Accordingly, the output of theadder 110becomes such an output as represented by equation (27), andlow-pass filter output YLPF(t) for the image sensor outputat position t of the pixel of interest can be obtained.YLPF(t) = S(t+2)/8+S(t+1)/4+S(t)/4+S(t-1)/4+S(t-2)/8
• While the process described above is applied to thelow-pass filter output for the image sensor output, the low-passfilter output of the pixel of the Jth color filter atposition t of the pixel of interest can be obtained by usinga two clock delay circuit instead of one clock delay circuit,because the Jth color filters are arranged at everyother pixel.
Embodiment 9
• Although the composition of the color video camera inembodiment 9 is the same as that of embodiment 2 (Fig.15),method of calculating the luminance signal in thearithmeticlogic unit 23 is different.
• Inembodiment 3, luminance signal is obtained by adding1 to each of the multiplier, divisor and dividend and subtracting1 from the result of calculation to minimize thecalculation error, as expressed by the equations (16) and(17). However, as thenumber 1 is LSB which has no significanteffect on the result of calculation, subtraction of 1at the end may be omitted for the simplification of thecircuit.Embodiment 9 is an example of such simplification.Luminance signal component Y(t) at the position of GC colorfilter of column t is calculated by the equation (28) below.Y(t) = (GC(t)+1) × ((YLPF(t)+1)/(GCLPF(t)+1))
• Luminance signal component Y(t) at position t of colorfilter of kind K (K is either MC, GY, MY or GC) is expressedby the equation (29) below.Y(t) = (K(t)+1) × ((YLPF(t)+1)/(KLPF(t)+1))
• Embodiment 9 is also capable of reducing the calculationerror similarly toembodiment 3.
Embodiment 10
• Fig.27 shows a block circuit diagram of the color videocamera inembodiment 10. In Fig.27, numerals which are thesame as those in Fig.15 indicate the identical portions.Numeral 29 represents a lookup table for logarithm andnumeral 30 represents a lookup table for power.
• The operation will now be described below. Luminancesignal component Y(t) in case the kind of color filter ofthe pixel at position t is K (K is either MC, GY, MY or GC)is given by the equation (29) below as described inembodiment9. Now apply logarithmic conversion with base x inequation (29) as shown in equation (30), where ^ representspower.Y(t) = x^log x {(K(t)+1) × ((YLPF(t)+1)/(KLPF(t)+1))}= x^{log x(K(t)+1) + log x (YLPF(t)+1)- log x (KLPF(t)+1)}
• If all coefficients used in equation (30) are given in8-bit numbers and the output of the lookup table forlogarithm29 is given with 10 bits, the capacity requirement is2560 bits. After calculating the logarithmic part by meansof the lookup table forlogarithm 29, addition and subtractionare carried out to calculate the power. Because thepower can be represented sufficiently with 11 bits, 8-bitoutput from the lookup table forpower 30 requires a capacityof 16384 bits. Consequently, calculation of the equation (29) can be done with a lookup table which has a capacity of18944 bits in total, making it possible to further simplifythe circuit composition.
• Althoughembodiment 10 is described in a case whereluminance signal is calculated from the equation ofembodiment9, it is a matter of course that the method of calculationwhich employs the lookup table for logarithm and thelookup table for power may be applied toembodiment 2.
Embodiment 11
• Fig.28 shows a block circuit diagram of the color videocamera inembodiment 11. In Fig.28, numerals which are thesame as those in Fig.5 indicate the identical portions andwill not be explained here. In Fig.28, numeral 31 representsa two-dimensional low-pass filter (LPF) for luminance signalY and numeral 32 represents an arithmetic logic unit. Fig.29shows the internal construction of thearithmetic logic unit32 which has ademultiplexer 33, adivider 34 and amultiplier35.
• The operation will now be described below. Basic operationis the same as that ofembodiment 1. The output of theimage sensor is fed from thefield memory 8 to the two-dimensionallow-pass filter 31, and output YLPF of the two-dimensionallow-pass filter as shown in Fig.30 is obtained.
• The arithmetic operation in thearithmetic logic unit 32 will be described below. Two-dimensional low-pass filteroutput of Y signal is fed to thedivider 34 as the dividend.Thedemultiplexer 33 switches to the two-dimensional low-passfilter output of MC if the pixel of interest is MCpixel, to the two-dimensional low-pass filter output of GYif the pixel of interest is GY pixel, to the two-dimensionallow-pass filter output of MY if the pixel ofinterest is MY pixel, or to the two-dimensional low-passfilter output of GC if the pixel of interest is GC pixel,with the output being fed to thedivider 34 as the divisor.The output of thisdivider 34 is fed to themultiplier 35 asthe input. The output signal of the pixel of interest isalso fed to themultiplier 35 as the input. Thus the outputof themultiplier 35 is obtained as the luminance signal Yof the pixel of interest.
• Embodiment 11 is an example of using two-dimensionallow-pass filters instead of one-dimensional low-pass filtersinembodiment 2. In Fig.6, luminance signal component Y(s,t)at the position of color filter GC of row s, column t iscalculated by equation (31).Y(s,t) = GC(s,t) × (YLPF(s,t)/GCLPF(s,t))
• Luminance signal component Y(s,t) at the position (s,t)of color filter of kind K (K is either MC, GY, MY or GC) isgiven by the equation (32) below, where (s,t) represents thecoordinates of thefield memory 8.Y(s,t) = K(s,t) × (YLPF(s,t)/KLPF(s,t))
• Calculation of Y signal in this embodiment is basicallythe same as that ofembodiment 2, and is capable of eliminatingthe modulated component of the color signal withoutreducing the harmonics of the luminance signal.
Embodiment 12
• Although the composition of the color video camera inembodiment 12 is the same as that of embodiment 11 (Fig.28),method of calculating Y signal in thearithmetic logic unit32 is different.
• The arithmetic operation in thearithmetic logic unit32 will be described below. In Fig.6, if each signal is madeup of eight bits to represent the hue in 256 steps, forexample, andvalue 1 of LSB is employed as a constant,luminance signal component Y(s,t) at the position of colorfilter GC of row s, column t is calculated by equation (33)as shown below.Y(s,t) = (GC(s,t)+1) × ((YLPF(s,t)+1)/GCLPF(s,t)+1))- 1
• Luminance signal component Y(s,t) at position (s,t) ofcolor filter of kind K (K is either MC, GY, MY or GC) isgiven by the equation (34) below.Y(s,t) = (K(s,t)+1) × ((YLPF(s,t)+1)/(KLPF(s,t)+1))- 1
Embodiment 13
• Fig.31 shows a block circuit diagram of the color videocamera inembodiment 13. In Fig.31, symbols which are thesame as those in Fig.28 indicate the identical portions andnumeral 27 represents a comparator which resembles that ofFig.22 (embodiment 4).
• The operation will now be described below. As in thecase ofembodiment 4, the difference between output signalsof pixels of the same kind which are fed to thecomparator27 from thefield memory 8 is compared to a particularthreshold to determine whether the spatial frequency of theimage is high or low. In a portion of high spatial frequency,luminance signal is calculated similarly toembodiment11 and, in a portion of low spatial frequency, luminancesignal is calculated from the weighted averaging value ofthe outputs of the pixels of N kinds in the vicinity of thepixel of interest.
• Thearithmetic logic unit 32 operates similarly to thatinembodiment 11 in the portion of high spatial frequency.In a portion of low spatial frequency, for example, luminancesignal component Y(s,t) at the position of colorfilter GC of row s, column t in Fig.6 is calculated byequation (35).Y(s,t) = MC(s-1,t-1)/16+GY(s-1,t)/8+MC(s-1,t+1)/16+MY(s,t-1)/8+GC(s,t)/4+MY(s,t+1)/8+MC(s+1, t-1)/16+GY(s+1,t)/8+MC(s+1,t+1)/16
• Assuming the position of the pixel of interest as(s,t), kind of color filter of the pixel of interest as J (Jis either MC, GY, MY or GC), kind of the color filter of theright and left adjacent pixels of the pixel of interest as K(K is either MC, GY, MY or GC), kind of the color filter ofthe upper and lower adjacent pixels of the pixel of interestas L (L is either MC, GY, MY or GC), and the kind of thecolor filter of the diagonally adjacent pixels of the pixelof interest as M (M is either MC, GY, MY or GC), then luminancesignal component Y(s,t) is calculated by the equation(36) below.Y(s,t) = M(s-1,t-1)/16+L(s-1,t)/8+M(s-1,t+1)/16+K(s,t-1)/8+J(s,t)/4+K(s,t+1)/8+M(s+1,t-1)/16+L(s+1,t)/8+M(s+1,t+1)/16
Embodiment 14
• Although the composition of the color video camera inembodiment 14 is the same as that of embodiment 13 (Fig.31),signal processing operations in thearithmetic logic unit 32and in thecomparator 27 are different.
• Inembodiment 14, difference between the output signalsof right and left pixels or between the output signals ofthe upper and lower pixels of the pixel of interest is compared to a particular threshold. And it is determinedthat the portion has high spatial frequency if the differencebetween the output signals is greater than the threshold,and accordingly the operation inembodiment 11 iscarried out, and, it is determined that the portion has lowspatial frequency if the difference between the outputsignals is less than the threshold, and accordingly theoperation inembodiment 13 is carried out.
• For example, the results of the calculations describedbelow are compared to the threshold at the position of thecolor filter of GC at row s, column t in Fig.6, to selectone of the methods of generating luminance signal.| MY(s,t-1) - MY(s,t+1) || GY(s-1,t) - GY(s+1,t) |
• Assuming the position of the pixel of interest as(s,t), the kind of the color filter of right and left adjacentpixels of the pixel of interest as J (J is either MC,GY, MY or GC), and the kind of the upper and lower adjacentpixels of the color pixel of interest as K (K is either MC,GY, MY or GC), the results of the calculations below arecompared to the threshold to select one of the methods ofgenerating luminance signal.| J(s,t-1) - J(s,t+1) || K(s-1,t) - K(s+1,t) |
Embodiment 15
• Although the composition of the color video camera inembodiment 15 is the same as that of embodiment 13 (Fig.31),signal processing operations in thearithmetic logic unit 32and in thecomparator 27 are different.
• Inembodiment 15, difference between the output signalsof the pixels of the same kind of spectral response characteristicin the vicinity of the pixel of interest is comparedto a particular threshold. And it is determined thatthe portion has high spatial frequency if the differencebetween the output signals is greater than the threshold,and the operation ofembodiment 11 is carried out, and, itis determined that the portion has low spatial frequency ifthe difference between the output signals is less than thethreshold, and the operation ofembodiment 13 is carriedout.
• For example, assuming that the color filter at theposition of coordinate (s,t) of the pixel of interest is GC,the color filter at the positions of the right and leftadjacent pixels of the pixel of interest is MY, the colorfilter at the positions of the upper and lower adjacentpixels of the pixel of interest is GY, and the color filterat the positions of the diagonally adjacent pixels of thepixel of interest is MC as shown in Fig.6, then the resultsof the calculations below are compared to the threshold to select one of the methods of generating luminance signal.| MY(s,t-1) - MY(s,t+1) || GY(s-1,t) - GY(s+1,t)|| MC(s-1,t-1) - MC(s+1,t+1) || MC(s-1,t+1) - MC(s+1,t-1) |
• Assuming the position of the pixel of interest as(s,t), the kind of the color filter of the right and leftadjacent pixels of the pixel of interest as J (J is eitherMC, GY, MY or GC), the kind of the color filter of the upperand lower adjacent pixels of the color pixel of interest asK (K is either MC, GY, MY or GC), and the kind of the colorfilter of the diagonally adjacent pixels of the pixel ofinterest as L (L is either MC, GY, MY or GC), then theresults of the calculations below are compared to thethreshold to select one of the methods of generating luminancesignal.| J(s,t-1) - J(s,t+1) || K(s-1,t) - K(s+1,t)|| L(s-1,t-1) - L(s+1,t+1) || L(s-1,t+1) - L(s+1,t-1) |
Embodiment 16
• Although the composition of the color video camera inembodiment 16 is the same as that of embodiment 13 (Fig.31),signal processing operations in thearithmetic logic unit 32 and in thecomparator 27 are different.
• Inembodiment 16, difference between the output signalsof the diagonally adjacent pixels interposing the pixel ofinterest is compared to a particular threshold. And it isdetermined that the portion has high spatial frequency ifthe difference between the output signals is greater thanthe threshold, and the operation inembodiment 11 is carriedout, and, it is determined that the portion has low spatialfrequency if the difference between the output signals isless than the threshold, and the operation inembodiment 13is carried out.
• For example, the results of the calculations shownbelow are compared to the threshold at the position of thecolor filter of GC at row s, column t in Fig.6, to selectone of the methods of generating luminance signal.| MC(s-1,t-1) - MC(s+1, t+1)|| MC(s+1,t-1) - MC(s-1,t+1) |
• Assuming the position of the pixel of interest as(s,t), the kind of the diagonally adjacent pixels of thecolor filter of the pixel of interest as J (J is either MC,GY, MY or GC), the results of the calculations shown beloware compared to the threshold to select one of the methodsof generating luminance signal.| J(s-1,t-1) - J(S+1,t+1) || J(s+1,t-1) - J(s-1,t+1) |
Embodiment 17
• Fig.32 shows a block circuit diagram of the color videocamera inembodiment 17. Symbols which are the same as thoseof Fig.28 indicate the identical portions and numeral 28represents a lookup table for division which is similar tothat of Fig.23 (embodiment 6).
• The operation in the lookup table fordivision 28 isthe same as that ofembodiment 6, and will not be describedhere.
Embodiment 18
• Application of the calculation method of embodiment 7with respect toembodiment 6 to theabove embodiment 17 istheembodiment 18. The operation inembodiment 18 is thesame as that of embodiment 7, and will be omitted.Embodiment 19
• The composition of the color video camera inembodiment19 is the same as that of embodiment 11 (Fig.28). Inembodiment19, a two-dimensional low-pass filter is used as adigital filter. The filter is made up of only bit shiftcircuits such as, for example, the number of horizontal tapsis set to 5, number of vertical taps being 3, the weightingsare 1/8, 1/4, 1/4, 1/4, 1/8 for the horizontal direction and1/4, 1/2, 1/4, for the vertical direction. Fig.33 shows thecomposition of the two-dimensional low-pass filter. In Fig.33, 201 and 202 are one horizontal period delay circuits(1HDLY), 204 is a 1-bit shift circuit (1B SHIFT), 203, 205,219 through 221, 224 through 226 and 229 through 231 are 2-bitshift circuits (2B SHIFT), 218, 222, 223, 227, 228, 232are 3-bit shift circuits (3B SHIFT), 206 through 217 are oneclock delay circuits (DLY) and 233 through 236 are adders.
• Assume that an image sensor output S(s+1,t+2) is fed tothe two-dimensional low-pass filter of such a composition.The output of the 2-bit shift circuit 203 is S(s+1,t+2)/4and the output of the 3-bit shift circuit 218 isS(s+1,t+2)/32. Output S(s+1,t+1)/4 of the image sensordelivered at the time one clock earlier is fed via the oneclock delay circuit 206 to the 2-bit shift circuit 219 toobtain an output S(s+1,t+1)/16. Similarly, outputsS(s+1,t)/16, S(s+1,t-1)/16 and S(s+1,t-2)/32 are obtainedfrom the 2-bit shift circuits 220, 221 and the 3-bit shiftcircuit 222. Accordingly, the output of theadder 233 isgiven as Y'LPF(s+1,t) expressed by equation (53).Y'LPF(s+1,t) = S(s+1,t+2)/32+S(s+1,t+1)/16+S(s+1,t)/16+ S(s+1,t-1)/16+S(s+1,t-2)/32
• Similarly, the output of theadder 234 is given asY'LPF(s,t) expressed by equation (54).Y'LPF(s,t) = S(s,t+2)/16+S(s,t+1)/8+S(s,t)/8+S(s,t-1)/8+S(s,t-2)/16
• Similarly, the output of theadder 235 is given as Y'LPF(s-1,t) expressed by equation (55).Y'LPF(s-1,t) = S(s-1,t+2)/32+S(s-1,t+1)/16+S(s-1,t)/16+S(s-1,t-1)/16+S(s-1,t-2)/32
• Accordingly, two-dimensional low-pass filter outputYLPF(s,t) of the image sensor output at the position (s,t)of the pixel of interest can be obtained in theadder 236 asshown by equation (56).YLPF(s,t) = Y'LPF(s+1,t)+Y'LPF(s,t)+Y'LPF(s-1,t)
• Although the described above are two-dimensional low-passfilter outputs of image sensor outputs, the two-dimensionallow-pass filters of the Jth color filters at theposition (s,t) of the pixel of interest are made of the Jthcolor filters which are arranged alternately at every otherpixel in both horizontal and vertical directions. Thereforeuse of a two clock delay circuit instead of a one clockdelay circuit, and a two horizontal period delay circuitinstead of a one horizontal period delay circuit will servethe purpose.
Embodiment 20
• Although the composition of the color video camera inembodiment 20 is the same as that of embodiment 11 (Fig.28),method of calculating the luminance signal in thearithmeticlogic unit 32 is different.
• In embodiment. 12, a luminance signal is obtained by adding 1 to each of the multiplier, divisor and dividend andsubtracting 1 from the result of calculation at the end tominimize the calculation error, as shown in equations (33)and (34). However, as thenumber 1 is LSB which has nosignificant effect on the result of calculation, subtractionof 1 at the end may be omitted for the simplification of thecircuit.Embodiment 20 is an example of such simplification.Luminance signal component Y(s,t) at the position of GCcolor filter of row s, column t is calculated by the equation(57) below.Y(s,t) = (GC(s,t)+1) × ((YLPF(s,t)+1)/(GCLPF(s,t)+1))
• Luminance signal component Y(s,t) at position (s,t) ofcolor filter of kind K (K is either MC, GY, MY or GC) isgiven by the equation (58) below.Y(s,t) = (K(s,t)+1) × ((YLPF(s,t)+1)/(KLPF(s,t)+1))
Embodiment 21
• Fig.34 shows a block circuit diagram of the color videocamera inembodiment 21. In Fig.34, symbols which are thesame as those in Fig.28 indicate the identical portions.Numerals 29, 30 represent a lookup table for logarithm and alookup table for power similar to those shown in Fig.27(embodiment 10).
• The operation will now be described below. Luminancesignal component Y(s,t) in case the kind of color filter ofthe pixel at position (s,t) is K (K is either MC, GY, MY orGC) is given, for example, by the equation (58) as describedinembodiment 20. Now logarithmic conversion with base x isapplied to equation (58) as shown in equation (59), where ^represents power.Y(s,t) = x^ log x {(K(s,t)+1) × ((YLPF(s,t)+1)/(KLPF(s,t)+1))}= x^ {log x (K(s,t)+1) + log x (YLPF(s,t)+1)- log x (KLPF(s,t)+1)}
• Inembodiment 21, arithmetic operation can be madeusing lookup tables of small capacity similarly toembodiment10 described before. Although the above description isfor the case of calculating a luminance signal based on theequation ofembodiment 20, it goes without saying that thecalculation by means of a lookup table for logarithm and alookup table for power may be applied toembodiment 11.
Embodiment 22
• Fig.35 shows a block circuit diagram of the color videocamera inembodiment 22. In Fig.35, numeral 51 represents alens, numeral 52 represents a refracting prism,numerals 53,54, 55 represent image sensors, numeral 56 represents a redsignal amplifier, numeral 57 represents a green signalamplifier, numeral. 58 represents a blue signal amplifier,numerals 59, 60, 61 represent A/D converters,numerals 62,63, 64 represent memories,numerals 65, 66, 67 representlow-pass filters (LPF), numeral 69 represents an adder,numeral 72 represents a demultiplexer, and numeral 75 representsan arithmetic logic unit. Fig.36 shows the internalcomposition of thearithmetic logic unit 75 which has demultiplexers80, 81, adivider 82 and amultiplier 83.
• The operation will now be described below. In Fig.35,incident ray on thelens 51 is decomposed into red, greenand blue by the refractingprism 52, with the light rays ofrespective colors forming images on theimage sensors 53,54, 55 which are formed by arranging photoelectric transducersoptically staggering from each other on a two-dimensionalplane. Each of theimage sensors 53, 54, 55mixes the signals of two upper and lower adjacent pixels togive one signal output. The output signals of theimagesensors 53, 54, 55 are amplified by thered signal amplifier56, thegreen signal amplifier 57 and theblue signal amplifier58, respectively, so that the ratio of the outputsignals thereof becomes, in the case of NTSC system,R:G:B=0.30:0.59:0.11. The amplified signals are convertedform analog to digital signals by the A/D converters 59, 60,61, respectively, to obtain R, G and B signals. R is the redpixel signal, G is the green pixel signal and B is the bluepixel signal. G signal is stored inmemory 62, R and B signals are mixed by theadder 69 and stored inmemory 63.Thedemultiplexer 72 switches alternately between G signaland R, B composite signal to produce a synthesized signal Y'which is stored inmemory 64. Fig.37, Fig.38 and Fig.39partly illustrate the configuration of G signal, R, B compositesignal and Y' signal written in thememories 62, 63,64. RB in the drawings represents the R, B composite signal.The signals shown in these drawings are smoothed by the low-passfilters 65, 66, 67. Fig.40, Fig.41 and Fig.42 show theoutputs of the low-pass filters 65, 66 and 67, respectively.LPF in the drawing is a symbol representing a low-passfilter output.
• The operation of thearithmetic logic unit 75 will bedescribed below. The low-pass filter output of the synthesizedsignal Y' is fed to thedivider 82 as the dividend.Thedemultiplexer 81 switches to the low-pass filter outputof G signal if the pixel of interest is a green pixel or tothe low-pass filter output of R, B composite signal if thepixel of interest is a red, blue pixel, and the outputthereof is fed to thedivider 82 as the divisor. The outputof thedivider 82 is supplied to themultiplier 83 as theinput. Thedemultiplexer 80 switches to the G signal if thepixel of interest is a green pixel or to the R, B compositesignal if the pixel of interest is a red, blue pixel, andthe output thereof, is fed to themultiplier 83. Thus the output of themultiplier 83 is obtained as Y signal at thepixel of interest.
• In Fig.37, luminance signal component Y (t) at theposition of green pixel of column t, for example, is calculatedby equation (60) below.Y(t) = G(t) × (Y'LPF(t)/GLPF(t))
• Luminance signal component Y (t) at the position t ofpixel of kind K (K is either G or RB) is calculated byequation (61) below. In this embodiment, letter t representsthe coordinate ofmemory 64 shown in Fig.39, or representsthe coordinate ofmemory 62 shown in Fig.37 in case thepixel of interest is G, or represents the coordinate ofmemory 63 shown in Fig.38 in case the pixel of interest isRB.Y(t) = K(t) × (Y'LPF(t)/KLPF(t))
Embodiment 23
• Fig.43 shows a block circuit diagram of the color videocamera inembodiment 23. In Fig.43, numerals which are thesame as those in Fig.35 indicate the identical portions andnumeral 85 represents an arithmetic logic unit with theinternal composition thereof being shown in Fig.44. Thearithmetic logic unit 85 has ademultiplexer 80, adivider82 and amultiplier 83.
• The operation will now be described below. Inembodiment 23, G signal and R, B composite signal from thedemultiplexer72 are fed to thearithmetic logic unit 85. Low-passfilter output of the synthesized signal Y' is fed tothedivider 82 as the dividend. Thedemultiplexer 80 switchesto the low-pass filter output of G signal if the pixel ofinterest is a green pixel, or to the low-pass filter outputof R, B composite signal if the pixel of interest is a red,blue pixel, and the output thereof is fed to thedivider 82as the divisor. The output of thedivider 82 is supplied tothemultiplier 83 as the input. Themultiplier 82 receives Gsignal as the input if the pixel of interest is a greenpixel, or R, B composite signal if the pixel of interest is.a red, blue pixel. As a result, output of themultiplier 83is obtained as Y signal at the pixel of interest. The restof the operation is the same as that ofembodiment 22.
• The principle of calculating Y signal in the above-mentionedembodiments 22 and 23 will now be described below.This method is based on an assumption that the color doesnot change significantly in a localized region. This impliesthat ratios of signals of different colors (G, RB) to Y'signal are approximately equal to each other in the localizedregion. Ratios between signals of different colors G,RB and Y' signal in a localized region are given by theratios between low-pass filter outputs of G, RB and low-passfilter output of Y'.
• At the position of G of column t as shown by equation(60), for example, Y(t) is obtained by multiplying G(t) bythe ratio of Y' and G in the localized region (the ratio ofthe low-pass filter output of Y' signal and low-pass filteroutput of G signal).
• In the conventional method of 3-chip CCD color camerabased on spatial offset, the purpose is set at improving theresolution. Although there arises no problem in the case ofsuch objects that have green signal and red-blue compositesignal in similar proportions, vertical lines appear in thecase of objects which have significantly different proportions.Vertical lines have been reduced by passing theluminance signal through a low-pass filter in the prior art,though it has a problem of causing attenuation of harmonicsin the luminance signal. According to this invention, Ysignal is obtained by multiplying G(t) by the ratio of Y'and G in the localized region (the ratio of the low-passfilter output of Y' signal and low-pass filter output of Gsignal), and consequently it is made possible to suppressthe generation of vertical lines without attenuating theharmonics of the luminance signal.
Embodiment 24
• Although the composition of the color video camera inembodiment 24 is the same as that shown in Fig.35, signalprocessing in thearithmetic logic unit 75 is different, and Y signal is calculated by employing 1 of LSB similarly toembodiment 3 described above. Specifically, Y(t) at theposition of G of column t is calculated by equation (62).Y(t) = (G(t)+1) × ((Y'LPF(t)+1)/(GLPF(t)+1))-1
• Luminance signal component Y(t) at the position t of apixel of kind K (K is either G or RB) is calculated byequation (63) below. Letter t represents the coordinate ofmemory 64 shown in Fig.39 in this embodiment, coordinate ofmemory 62 shown in Fig.37 in case the pixel of interest isG, or coordinate ofmemory 63 shown in Fig.38 in case thepixel of interest is RB.Y(t) = (K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1))- 1
Embodiment 25
• Fig.45 shows a block circuit diagram of the color videocamera inembodiment 25. In Fig.45, symbols which are thesame as those in Fig.35 indicate the identical portions, andnumeral 27 represents a comparator similar to that shown inFig.22 (embodiment 4).
• The operation will now be described below. The outputsignals of an appropriate pixel in the vicinity of the pixelof interest are supplied frommemories 62, 63 to thecomparator27. And it is determined that the image has a high spatial frequency if the difference between the outputsignals from pixels of the same kind is greater than aparticular threshold, and it is determined that the imagehas a low spatial frequency if the difference between theoutput signals is within the threshold. In a portion ofimage having a high spatial frequency, luminance signal iscalculated similarly toembodiment 22 and, in a portion oflow spatial frequency, luminance signal is calculated fromthe weighted averaging value of the outputs of G signal andRB signal.
• Thearithmetic logic unit 75 operates similarly toembodiment 22 in the portion of high spatial frequency. Inthe portion of low spatial frequency, luminance signalcomponent Y(t) at the position where there is no outputsignal of column t in Fig.38, for example, is calculated byequation (64) below.Y(t) = RB(t-1)/4+G(t)/2+RB(t+1)/4
• When the pixel of interest is located at position t,the kind of the pixel of interest is J (J is either G orRB), and the kind of the right and left horizontal pixels ofthe pixel of interest is K (K is either G or RB), luminancesignal component Y(t) is calculated by equation (65).Y(t) = K(t-1)/4+J(t)/2+K(t+1)/4
Embodiment 26
• Although the composition of the color video camera inembodiment 26 is the same as that of embodiment 25 (Fig.45),signal processing operations in thearithmetic logic unit 75and in thecomparator 27 are different.
• Inembodiment 26, difference the between the outputsignals of the right and left adjacent pixels of the pixelof interest is compared to a particular threshold. And it isdetermined that the portion has high spatial frequency ifthe difference between the output signals is greater thanthe threshold, and the operation ofembodiment 22 is carriedout, and, it is determined that the portion has low spatialfrequency if the difference between the output signals isless than the threshold, and the operation ofembodiment 25described above is carried out.
• For example, the result of the following calculation iscompared to the threshold at the position of G of column tin Fig.37 and at the position where there is no outputsignal of column t in Fig.38, to select one of the methodsof generating luminance signal.| RB(t-1) - RB(t+1) |
• Assuming the position of the pixel of interest as t,the kind of the right and left adjacent pixels of the pixelof interest as K (K is either G or RB), the result of thecalculation below is compared to the threshold to select oneof the methods of generating luminance signal.| K(t-1) - K(t+1) |
Embodiment 27
• Fig.46 shows a block circuit diagram of the color videocamera inembodiment 27. In Fig.46, symbols which are thesame as those in Fig.35 indicate the identical portions, andnumeral 28 represents a lookup table for division similar tothat shown in Fig.23 (embodiment 6).
• The operation of the lookup table fordivision 28 isthe same as that inembodiment 6, and description thereofwill be omitted here.
Embodiment 28
• An embodiment where the method of calculation in embodiment7 with respect toembodiment 6 is applied toembodiment27 described above isembodiment 28. The operation ofthe lookup table fordivision 28 inembodiment 28 is thesame as that in embodiment 7, and description thereof willbe omitted here.
Embodiment 29
• The composition of the color video camera inembodiment29 is the same as that of embodiment 22 (Fig.35). Inembodiment29, a one-dimensional low-pass filter is used as adigital filter which is made up of only bit shift circuitshaving weightings such as 1/8, 1/4. The construction of theone-dimensional low-pass filter is the same as that in embodiment 8 (Fig.26) and will not be described here.Embodiment 30
• Although the composition of the color video camera inembodiment 30 is the same as that of embodiment 22 (Fig.35),the method of calculating the luminance signal in thearithmeticlogic unit 75 is different.
• Inembodiment 24, the luminance signal is obtained byadding 1 to each of the multiplier, divisor and dividend andsubtracting 1 from the result of calculation at the end tominimize the calculation error, as shown in equations (62)and (63). However, as thenumber 1 is LSB which has nosignificant effect on the result of calculation, subtractionof 1 at the end may be omitted for the simplification of thecircuit.Embodiment 30 is an example of such simplification.Luminance signal component Y(t) at the position of G ofcolumn t is calculated by the equation (68) below.Y(t) = (G(t)+1) × ((Y'LPF(t)+1)/(GLPF(t)+1))
• Luminance signal component Y(t) at position t of pixelof kind K (K is either G or RB) is given by the equation(69) below.Y(t) = (K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1))
Embodiment 31
• Fig.47 shows a block circuit diagram of the color videocamera inembodiment 31. In Fig.47, symbols which are thesame as those in Fig.35 indicate the identical portions.Numerals 29, 30 represent a lookup table for logarithm and alookup table for power which are similar to those shown inFig.27 (embodiment 10).
• The operation will now be described below. Luminancesignal component Y(t) in case the kind of the pixel atposition t is K (K is either G or RB), for example, is givenby equation (69) as described inembodiment 30. Now logarithmicconversion with base x as shown in equation (70) isapplied to equation (69), where ^ represents power.Y(t) = x^log x {(K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1))}= x^ {log x (K(t)+1) + log x (Y'LPF(t)+1)- log x (KLPF(t)+1)}
• Inembodiment 31, calculation can be made by usinglookup tables of small capacity as inembodiment 10 describedbefore. Although the above description is for thecase of calculating luminance signal based on the equationofembodiment 30, it goes without saying that the calculationby means of a lookup table for logarithm and a lookuptable for power may be applied toembodiment 22.
Embodiment 32
• Fig.48 shows a block circuit diagram of the color video camera inembodiment 32. In Fig.48, numerals which are thesame as those in Fig.35 indicate the identical portions andwill not be described here. In Fig.48,numerals 92, 93, 94represent two-dimensional memories, andnumerals 95, 96, 97represent two-dimensional low-pass filters (LPF).
• The operation will now be described below. The basicoperation is the same as that ofembodiment 22. The configurationof G signal, R, B composite signal and Y' signalwritten in the two-dimensional memories 95, 96, 97 is partiallyillustrated in Fig.49, Fig.50 and Fig.51. The signalsshown in these drawings are smoothed by the two-dimensionallow-pass filters 95, 96, 97. Fig.52, Fig.53, Fig.54 show theoutputs of the two-dimensional low-pass filters 95, 96, 97.Letters LPF in the drawing indicate the low-pass filteroutput.
• Embodiment 32 is an example of using two-dimensionallow-pass filters instead of one-dimensional low-pass filtersofembodiment 22. In Fig.49, luminance signal componentY(s,t) at the position of green pixel of row s, column t,for example, is calculated by equation (71).Y(s,t) = G(s,t) × (Y'LPF(s,t)/GLPF(s,t))
• Luminance signal component Y(s,t) at the position (s,t)of a pixel of kind K (K is either G or RB) is given byequation (72) below, where (s,t) represents the coordinatesof the two-dimensional memory 94 shown in Fig.51, the coordinates of the two-dimensional memory 92 shown in Fig.49 incase the pixel of interest is G, and the coordinates of thetwo-dimensional memory 93 shown in Fig.50 in case the pixelof interest is RB.Y(s,t) = K(s,t) × (Y'LPF(s,t)/KLPF(s,t))
Embodiment 33
• Fig.55 shows a block circuit diagram of the color videocamera inembodiment 33. In Fig.55, numerals which are thesame as those in Fig.48 indicate the identical portions andwill not be described here. In Fig.55, thearithmetic logicunit 85 has the same composition as that of thearithmeticlogic unit 85 inembodiment 23 described previously(Fig.44). In thisembodiment 33, ademultiplexer 72 feeds Gsignal and R, B composite signal to thearithmetic logicunit 85. Inembodiment 33, two-dimensional low-pass filtersare used instead of one-dimensional low-pass filters ofembodiment 23.
• The principle of calculating the Y signal inembodiments32, 33 is basically the same as that inembodiments22, 23, and is capable of eliminating the modulated componentsof the color signal without reducing the harmonics ofthe luminance signal.
Embodiment 34
• Although the composition of the color video camera inembodiment 34 is the same as that of embodiment 32 (Fig.48),the method of calculating the Y signal in thearithmeticlogic unit 75 is different.
• The arithmetic operation in thearithmetic logic unit75 will be described below. If each signal is made up ofeight bits to represent the hue in 256 steps, for example,andvalue 1 of LSB is employed as a constant, the luminancesignal component Y(s,t) at the position of G of row s,column t is calculated by equation (73) as shown below.Y(s,t) = (G(s,t)+1) × ((Y'LPF(s,t)+1)/(GLPF(s,t)+1)) - 1
• The luminance signal component Y(s,t) at position (s,t)of pixel of kind K (K is either G or RB) is given by equation(74) below.Y(s,t) = (K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1)) - 1
Embodiment 35
• Fig.56 shows a block circuit diagram of the color videocamera inembodiment 35. In Fig.56, symbols which are thesame as those in Fig.48 indicate the identical portions andnumeral 27 represents a comparator which is similar to thatof Fig.22 (embodiment 4).
• The operation will now be described below. Differencebetween output signals of pixels of the same kind which are fed to thecomparator 27 from the two-dimensional memories92, 93 is compared to a particular threshold to determinewhether the spatial frequency of the image is high or low.In a portion of the image with high spatial frequency, theluminance signal is calculated similarly toembodiment 32and, in a portion of low spatial frequency, the luminancesignal is calculated from the weighted averaging value ofthe G signal and RB signal in the vicinity of the pixel ofinterest.
• Thearithmetic logic unit 75 operates similarly to thatinembodiment 32 in a portion of high spatial frequency. Ina portion of low spatial frequency, the luminance signalcomponent Y(s,t) at the position of G of row s, column t inFig.49, and at a position where there is no output signal ofrow s, column t in Fig.50, for example, is calculated byequation (75).Y(s,t) = RB(s-1,t-1)/16+G(s-1,t)/8+RB(s-1,t+1)/16+RB(s,t-1)/8+G(s,t)/4+RB(s,t+1)/8+RB(s+1,t-1)/16+G(s+1,t)/8+RB(s+1,t+1)/16
• Assuming the position of the pixel of interest as(s,t), kind of the pixel of interest as J (J is either G orRB), kind of the right and left adjacent pixels of the pixelof interest as K (K is either G or RB), then the luminance signal component Y(s,t) is calculated by equation (76)below.Y(s,t) = K(s-1,t-1)/16+J(s-1,t)/8+K(s-1,t+1)/16+K(s,t-1)/8+J(s,t)/4+K(s,t+1)/8+K(s+1,t-1)/16+J(s+1,t)/8+K(s+1,t+1)/16
Embodiment 36
• Although the composition of the color video camera inembodiment 36 is the same as that of embodiment 35 (Fig.56),the signal processing operations in thearithmetic logicunit 75 and in thecomparator 27 are different.
• In embodiment 36, the difference between the outputsignals of the right and left pixels, or between the outputsignals of the upper and lower pixels, of the pixel ofinterest is compared to a particular threshold. And it isdetermined that the portion has high spatial frequency ifthe difference between the output signals is greater thanthe threshold, and the operation inembodiment 32 is carriedout, and, it is determined that the portion has low spatialfrequency if the difference between the output signals isless than the threshold, and accordingly the operation inembodiment 35 described above is carried out.
• For example, result of the calculation described belowis compared to a particular threshold at the position of G at row s, column t in Fig.49, and at the position wherethere is no output signal at row s, column t in Fig.50,thereby to select one of the methods of generating theluminance signal.| RB(s,t-1) - RB(s,t+1) || G(s-1,t) - G(s+1,t) |
• Assuming the position of the pixel of interest as(s,t), the kind of the pixel of interest as J (J is either Gor RB), and the kind of the right and left adjacent pixelsof the pixel of interest as K (K is either G or RB), thenthe results of the calculations below are compared to aparticular threshold to select one of the two methods ofgenerating the luminance signal.| K(s,t-1) - K(s,t+1)|| J(s-1,t) - J(s+1,t) |
Embodiment 37
• Although the composition of the color video camera inembodiment 37 is the same as that of embodiment 35 (Fig.56),the signal processing operations in thearithmetic logicunit 75 and in thecomparator 27 are different.
• In embodiment 37, the difference between the outputs ofthe pixels of the same kind in the vicinity of the pixel ofinterest is compared to a particular threshold. And it isdetermined that the portion has high spatial frequency if the difference is greater than the threshold, and theoperation ofembodiment 32 is carried out, and, it is determinedthat the portion has low spatial frequency if thedifference is less than the threshold, and accordingly theoperation ofembodiment 35 described previously is carriedout.
• For example, the results of the calculations below arecompared to a particular threshold at the position of G ofrow s, column t in Fig.49 and at a position where there isno output signal of row s, column t in Fig.50, to select oneof the two methods of generating the luminance signal.| RB(s,t-1) - RB(s,t+1) || G(s-1,t) - G(s+1,t) || RB(s-1,t-1) - RB(s+1,t+1) || RB(s-1,t+1) - RB(s+1,t-1) |
• Assuming that the position of the pixel of interest is(s, t), the kind of the pixel of interest as J (J is eitherG or RB), and the kind of the right and left adjacent pixelsof the pixel of interest as K (K is either G or RB), theresults of the calculations below are compared to a particularthreshold to select one of the two methods of generatingthe luminance signal.| K(s,t-1) - K(s,t+1) || J(s-1,t) - J(s+1,t) || K(s-1,t-1) - K(s+1,t+1) || K(s-1,t+1) - K(s+1,t-1) |
Embodiment 38
• Although the composition of the color video camera inembodiment 38 is the same as that of embodiment 35 (Fig.56),the signal processing operations in thearithmetic logicunit 75 and in thecomparator 27 are different.
• In embodiment 38, the difference between the outputsignals of the diagonally adjacent pixels interposing thepixel of interest is compared to a particular threshold. Andit is determined that the portion has high spatial frequencyif the difference between the output signals is greater thanthe threshold, and the operation inembodiment 32 is carriedout, and, it is determined that the portion has low spatialfrequency if the difference between the output signals isless than the threshold, and the operation inembodiment 35described previously is carried out.
• For example, results of the calculations shown beloware compared to the threshold at the position of G at row s,column t in Fig.49, and at a position where there is nooutput signal of row s, column t in Fig.50, thereby toselect one of the two methods of generating the luminancesignal accordingly.| RB(s-1,t-1) - RB(s+1,t+1) || RB(s+1,t-1) - RB(s-1,t+1) |
• Assuming the position of the pixel of interest as(s,t), the kind of the diagonally adjacent pixels of thepixel of interest as J (J is either G or RB), the results ofthe calculations shown below are compared to the threshold,thereby to select one of the two methods of generating theluminance signal.| J(s-1,t-1) - J(S+1,t+1) || J(s+1,t-1) - J(s-1,t+1) |
Embodiment 39
• Fig.57 shows a block circuit diagram of the color videocamera in embodiment 39. In Fig.57, symbols which are thesame as those of Fig.48 indicate the identical portions, andnumeral 28 represents a lookup table for division which issimilar to that of Fig.23 (embodiment 6).
• The operation in the lookup table fordivision 28 isthe same as that ofembodiment 6, and will not be describedhere.
Embodiment 40
• Application of the method of calculation in embodiment7 with respect toembodiment 6 to the above embodiment 39 isthis embodiment 40. The operation of the lookup table fordivision 28 in embodiment 40 is the same as that of embodiment7, and will not be described here.
Embodiment 41
• The composition of the color video camera in embodiment41 is the same as that of embodiment 32 (Fig.48). In embodiment41, a two-dimensional low-pass filter made up of onlybit shift circuits of weightings such as 1/8, 1/4, 1/2 isused as a digital filter. The composition of the two-dimensionallow-pass filter is similar to that of embodiment19 (Fig.33) where output Y'LPF(s, t) is obtained from thetwo-dimensional low-pass filter in response to the synthesizedsignal Y', and description thereof will be omittedhere. It should be noted here, however, that use of a twoclock delay circuit instead of a one clock delay circuitserves the purpose, because the two-dimensional low-passfilter for the Kth pixel at the position (s, t) of the pixelof interest is made by arranging the Kth pixels alternatelyevery other pixel in the horizontal direction.
Embodiment 42
• Although the composition of the color video camera inembodiment 42 is the same as that of embodiment 32 (Fig.48),the method of calculating the luminance signal in thearithmeticlogic unit 75 is different.
• Inembodiment 34, the luminance signal is obtained byadding 1 to each of the multiplier, divisor and dividend andsubtracting 1 from the result of calculation at the end tominimize the calculation error, as shown in equations (73) and (74). However, because thenumber 1 is LSB which has nosignificant effect on the result of calculation, subtractionof 1 at the end may be omitted for the simplification of thecircuit.Embodiment 42 is an example of such simplification.The luminance signal component Y(s,t) at the position of Gof row s, column t is calculated by equation (93) below.Y(s,t) = (G(s,t)+1) × ((Y'LPF(s,t)+1)/(GLPF(s,t)+1))
• The luminance signal component Y(s,t) at position (s,t)of pixel of kind K (K is either G or RB) is given by equation(94) below.Y(s,t) = (K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1))
Embodiment 43
• Fig.58 shows a block circuit diagram of the color videocamera inembodiment 43. In Fig.58, symbols which are thesame as those in Fig.48 indicate the identical portions.Numerals 29, 30 represent a lookup table for logarithm and alookup table for power similar to those shown in Fig.27(embodiment 10).
• The operation will now be described below. The calculationof the luminance signal component Y(s,t) in case thekind of the pixel at position (s,t) is K (K is either G orRB) is given, for example, by the equation (94) as described inembodiment 42. To this equation (94), the logarithmicconversion with base x is applied as shown in equation (95),where ^ represents power.Y(s,t) = X^log x{(K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1))}= X^{log x(K(s,t)+1) + log x(Y'LPF(s,t)+1)- log x(KLPF(s,t)+1)}
• In thisembodiment 43, the arithmetic operation can bemade using lookup tables of small capacity similarly toembodiment 10 described before. Although the above descriptionis for the case of calculating a luminance signal basedon the equation ofembodiment 42, the calculation by meansof a lookup table for logarithm and a lookup table for powermay be applied toembodiment 32.
Embodiment 44
• Fig.59 shows a block circuit diagram of the color videocamera in embodiment 44. In Fig.59, numerals which are thesame as those in Fig.15 indicate the identical portions. InFig.59, numeral 45 represents a low-pass filter (LPF),numeral 46 represents a band-pass filter (BPF) and numeral47 represents an adder.
• The operation will now be described below. The basicoperation from thelens 1 to thearithmetic logic unit 23 isthe same as that ofembodiment 2. The output of theimage sensor 2 is fed to the low-pass filter 45 to obtain YLsignal as the output. The YL signal is similar to the luminancesignal obtained with the prior art (see Fig.2).
• The operation of thearithmetic logic unit 23 will bedescribed below. When the low-pass filter output of Y'signal is fed to thedivider 25 as the dividend (Fig.16),the arithmetic operation similar to that inembodiment 2 isperformed to obtain the output of themultiplier 26 as theYH' signal of the pixel of interest.
• In Fig.17, YH' signal at the position of color filterGC of column t, for example, is calculated by equation (96)below.YH'(t) = GC(t) × (Y'LPF(t)/GCLPF(t))
• YH' signal at the position t of the color filter ofkind K (K is either MC, GY, MY or GC) is calculated byequation (97) below, similarly toembodiment 2.YH'(t) = K(t) × (Y'LPF(t)/KLPF(t))
• The aperture correction is carried out by taking harmonicscomponent YH from the YH' signal by means of theband-pass filter 46, combining the harmonics component YHand the YL signal in theadder 47 and thereby obtaining theluminance signal Y.
• The principle of calculating the YH' signal in thisembodiment 44 is basically the same as the calculation ofthe Y signal inembodiment 2. Calculating the YH' signal in this way makes it possible to eliminate the modulated componentsof the color signal without reducing the harmonics ofthe luminance signal. Consequently, the aperture correctionwithout unnatural enhancement is made possible by taking theharmonics component YH from the YH' signal and mixing itwith the YL signal.
Embodiment 45
• Although the composition of the color video camera inembodiment 45 is the same as that of embodiment 44 (Fig.59),the method of calculating the YH' signal in thearithmeticlogic unit 23 is different.
• The arithmetic operation in thearithmetic logic unit23 will be described below. If each signal is made up ofeight bits to represent the hue in 256 steps, for example,andvalue 1 of LSB is employed as a constant, YH' signal atthe position of color filter GC of column t is calculated byequation (98) as shown below.YH'(t) = (GC(t)+1) × ((Y'LPF(t)+1)/(GCLPF(t)+1)) - 1
• The calculation of YH' signal at the position t ofcolor filter of kind K (K is either MC, GY, MY or GC) isgiven by equation (99).YH'(t) = (K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1)) - 1
Embodiment 46
• Fig.60 shows a block circuit diagram of the color videocamera inembodiment 46. In Fig.60, symbols which are thesame as those in Fig.59 indicate the identical portions, andnumeral 27 represents a comparator similar to that shown inFig.22 (embodiment 4).
• The operation will now be described below. The outputsignals of appropriate pixels in the vicinity of the pixelof interest are supplied from thefield memory 8 to thecomparator 27. In a portion where the image has a highspatial frequency, YH' signal is calculated similarly toembodiment 44 and, in a portion of a low spatial frequency,YH' signal is calculated from the weighted averaging valueof the outputs of pixels of N kinds in the vicinity of thepixel of interest.
• Thearithmetic logic unit 23 operates similarly toembodiment 44 in the portion of high spatial frequency. Inthe portion of low spatial frequency, YH' signal at theposition of color filter of GC of column t in Fig.17, and atthe position where there is no output signal of column t inFig.18, for example, is calculated by equation (100) below,similarly toembodiment 4.YH'(t) = MY(t-1)/4+GC(t)/2+MY(t+1)/4
• The YH' signal, when the pixel of interest is locatedat position t, kind of the color filter of the pixel of interest is J and the kind of the color filters of the rightand left adjacent pixels of the pixel of interest is K, iscalculated by equation (101) below similarly toembodiment4.YH'(t) = K(t-1)/4+J(t)/2+K(t+1)/4
Embodiment 47
• Although the composition of the color video camera inembodiment 47 is the same as that of embodiment 46 (Fig.60),the signal processing operations in thearithmetic logicunit 23 and in thecomparator 27 are different.
• Inembodiment 47, the difference between the outputsignals of the right and left adjacent pixels of the pixelof interest is compared to a particular threshold. And it isdetermined that the portion has a high spatial frequency ifthe difference between the output signals is greater thanthe threshold, and the operation in embodiment 44 is carriedout, and, it is determined that the portion has a low spatialfrequency if the difference between the output signalsis less than the threshold, and accordingly the operation inembodiment 46 is carried out.
• Results of the calculations by the same equations (20),(21) as inembodiment 5 are compared to a particular threshold,and the method of calculating the YH' signal is selectedaccording to the result of comparison.
Embodiment 48
• Fig.61 shows a block circuit diagram of the color videocamera in embodiment 48. In Fig.61, symbols which are thesame as those in Fig.59 indicate the identical portions, andnumeral 28 represents a lookup table for division similar tothat shown in Fig.23 (embodiment 6).
• The operation of the lookup table fordivision 28 isthe same as that inembodiment 6, and will not be describedhere.
Embodiment 49
• Application of the calculation method of embodiment 7with respect toembodiment 6 to the above embodiment 48 isthis embodiment 49. The operation of the lookup table fordivision 28 in embodiment 49 is the same as that of embodiment7, and will not be described here.
Embodiment 50
• The composition of the color video camera in embodiment50 is the same as that of embodiment 44 (Fig.59). In embodiment50, a one-dimensional low-pass filter made up of onlybit shift circuits of weightings such as 1/8, 1/4 is used asa digital filter. The composition of the one-dimensionallow-pass filter is similar to that of embodiment 8 (Fig.26),and description thereof will be omitted here.
Embodiment 51
• Although the composition of the color video camera inembodiment 51 is the same as that of embodiment 44 (Fig.59),the method of calculating the YH' signal in thearithmeticlogic unit 23 is different.
• Inembodiment 45, YH' signal is obtained by adding 1 toeach of the multiplier, divisor and dividend and subtracting1 from the result of calculation at the end to minimize thecalculation error, as shown in equations (98) and (99).However, because thenumber 1 is LSB which has no significanteffect on the result of calculation, subtraction of 1at the end may be omitted for the simplification of thecircuit.Embodiment 51 is an example of such simplification,where YH' signal at the position of color filter GC ofcolumn t is calculated by equation (102) below.YH'(t) = (GC(t)+1) × ((Y'LPF(t)+1)/(GCLPF(t)+1))
• The YH' signal at the position t of color filter ofkind K is given by equation (103) below.YH'(t) = (K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1))
Embodiment 52
• Fig.62 shows a block circuit diagram of the color videocamera inembodiment 52. In Fig.62, symbols which are thesame as those in Fig.59 indicate the identical portions, andnumerals 29, 30 represent a lookup table for logarithm and a lookup table for power which are similar to those shown inFig.27 (embodiment 10).
• The operation will now be described below. The calculationof the YH' signal in case the kind of the color filterat position t is K, for example, is given by equation (103)as described inembodiment 51. Logarithmic conversion withbase x as shown in equation (104) is applied to equation(103), where ^ represents power.YH'(t) = X^log x{(K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1))}= X^ {log x(K(t)+1) + log x(Y'LPF(t)+1)- log x(KLPF(t)+1)}
• In thisembodiment 52, too, the calculation can becarried out with lookup tables of small capacity as in thecase ofembodiment 10 described previously. Although theabove description is for the case of calculating YH' signalbased on the equation ofembodiment 51, it goes withoutsaying that the calculation by means of a lookup table forlogarithm and a lookup table for power may be applied toembodiment 44.
Embodiment 53
• Fig.63 shows a block circuit diagram of the color videocamera inembodiment 53. In Fig.63, numerals which are thesame as those of Fig.28 indicate the identical portions, andwill not be described here.
• The operation will now be described below.Embodiment53 is an example of using two-dimensional low-pass filtersinstead of one-dimensional low-pass filters in embodiment44. In Fig.6, YH' signal at the position of the color filterGC of row s, column t, for example, is calculated by equation(105).YH'(s,t) = GC(s,t) × (Y'LPF(s,t)/GCLPF(s,t))
• The calculation of the YH' signal at the position (s,t)of color filter of kind K is given by equation (106) below.YH'(s,t) = K(s,t) × (Y'LPF(s,t)/KLPF(s,t))
• The principle of calculating the YH' signal in thisembodiment is basically the same as that in embodiment 44.Similarly to the case of embodiment 44, it is possible toeliminate the modulated components of the color signalwithout reducing the harmonics of the luminance signal,thereby enabling to carry out aperture correction withoutunnatural enhancement.
Embodiment 54
• Although the composition of the color video camera inembodiment 54 is the same as that of embodiment 53 (Fig.63),the method of calculating the Y signal in thearithmeticlogic unit 32 is different.
• The arithmetic operation in thearithmetic logic unit32 will be described below. If each signal is made up ofeight bits to represent the hue in 256 steps, for example, andvalue 1 of LSB is employed as a constant, YH' signal atthe position of color filter GC of row s, column t is calculatedby equation (107) as shown below.YH'(s,t) = (GC(s,t)+1) × ((Y'LPF(s,t)+1)/(GCLPF(S,T)+1)) - 1
• The calculation of the YH' signal at the position (s,t)of color filter of kind K is given by equation (108) below.YH'(s,t) = (K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1)) - 1
Embodiment 55
• Fig.64 shows a block circuit diagram of the color videocamera inembodiment 55. In Fig.64, symbols which are thesame as those in Fig.63 indicate the identical portions andnumeral 27 represents a comparator which is similar to thatof Fig.22 (embodiment 4).
• The operation will now be described below. The outputsignals of appropriate pixels around the pixel of interestare fed to thecomparator 27 from thefield memory 8. In aportion of the image with a high spatial frequency, YH'signal is calculated similarly toembodiment 53 and, in aportion of low spatial frequency, YH' signal is calculatedfrom the weighted averaging value of the outputs of pixelsof N kinds around the pixel of interest.
• Thearithmetic logic unit 32 operates similarly to that inembodiment 53 in the portion of high spatial frequency.In a portion of low spatial frequency, YH' signal at positionof color filter GC of row s, column t in Fig.6, forexample, is calculated by equation (109).YH'(s,t) = MC(s-1,t-1)/16+GY(s-1,t)/8+MC(s-1,t+1)/16+MY(s,t-1)/8+GC(s,t)/4+MY(s,t+1)/8+MC(s+1, t-1)/16+GY(s+1, t)/8+MC(s+1,t+1)/16
• Assuming the position of the pixel of interest as (s,t), kind of color filter of the pixel of interest as J, kindof the color filters of the right and left adjacent pixelsof the pixel of interest as K, kind of the color filters ofthe upper and lower adjacent pixels of the pixel of interestas L, kind of the color filters of the diagonally adjacentpixels of the pixel of interest as M, then YH' signal iscalculated by equation (110) below.YH'(s,t) = M(s-1,t-1)/16+L(s-1,t)/8+M(s-1,t+1)/16+K(s,t-1)/8+J(s,t)/4+K(s,t+1)/8+M(s+1,t-1)/16+L(s+1,t)/8+M(s+1,t+1)/16
Embodiment 56
• Although the composition of the color video camera inembodiment 56 is the same as that of embodiment 55 (Fig.64),the signal processing operations in thearithmetic logic unit 32 and in thecomparator 27 are different.
• Inembodiment 56, the difference between the outputsignals of the right and left adjacent pixels of the pixelof interest or the difference between the output signals ofthe upper and lower adjacent pixels of the pixel of interestis compared to a particular threshold. And it is determinedthat the portion has high spatial frequency if the differencebetween the output signals is greater than the threshold,and the operation ofembodiment 53 is carried out, andit is determined that the portion has low spatial frequencyif the difference between the output signals is less thanthe threshold, and accordingly the operation ofembodiment55 described above is carried out.
• Results of calculation of the same equations (37)through (40) as inembodiment 14 described previously arecompared to a particular threshold, and the method of calculatingthe YH' signal is selected according to the result ofcomparison.
Embodiment 57
• Although the composition of the color video camera inembodiment 57 is the same as that of embodiment 55 (Fig.64),the signal processing operations in thearithmetic logicunit 32 and in thecomparator 27 are different.
• Inembodiment 57, the difference between pixels of thesame spectral response characteristics around the pixel of interest is compared to a particular threshold. And it isdetermined that the portion has a high spatial frequency ifthe difference is greater than the threshold, and the operationofembodiment 53 is carried out, and, it is determinedthat the portion has a low spatial frequency if the differenceis less than the threshold, and the operation ofembodiment55 described above is carried out.
• Results of calculation of the same equations (41)through (48) as inembodiment 15 described previously arecompared to a particular threshold, and the method of calculatingthe YH' signal is selected according to the result ofcomparison.
Embodiment 58
• Although the composition of the color video camera inembodiment 58 is the same as that of embodiment 55 (Fig.64),the signal processing operations in thearithmetic logicunit 32 and in thecomparator 27 are different.
• Inembodiment 58, the difference between the outputsignals of diagonally adjacent pixels interposing the pixelof interest is compared to a particular threshold. And it isdetermined that the portion has a high spatial frequency ifthe difference between the output signals is greater thanthe threshold, and the operation ofembodiment 53 is carriedout, and, it is determined that the portion has a low spatial frequency if the difference between the output signalsis less than the threshold, and accordingly the operation ofembodiment 55 described previously is carried out.
• Results of calculation of the same equations (49)through (52) as inembodiment 16 described previously arecompared to a particular threshold, and the method of calculatingthe YH' signal is selected according to the result ofcomparison.
Embodiment 59
• Fig.65 shows a block circuit diagram of the color videocamera inembodiment 59. In Fig.65, symbols which are thesame as those in Fig.63 indicate the identical portions, andnumeral 28 represents a lookup table for division similar tothat shown in Fig.23 (embodiment 6).
• The operation of the lookup table fordivision 28 isthe same as that inembodiment 6, and will not be describedhere.
Embodiment 60
• Application of the calculation method of embodiment 7with respect toembodiment 6 to theabove embodiment 59 isthisembodiment 60. The operation of the lookup table fordivision 28 inembodiment 60 is the same as that of embodiment7, and will not be described here.
Embodiment 61
• The composition of the color video camera in embodiment,61 is the same as that of embodiment 53 (Fig.63). Inembodiment61, a two-dimensional low-pass filter made up of onlybit shift circuits of weightings such as 1/8, 1/4, 1/2 isused as a digital filter. The composition of the two-dimensionallow-pass filter is similar to that of embodiment19 (Fig.33), and the description thereof will be omitted.
Embodiment 62
• Although the composition of the color video camera inembodiment 62 is the same as that of embodiment 53 (Fig.63),the method of calculating the YH' signal in thearithmeticlogic unit 32 is different.
• Inembodiment 54, the luminance signal is obtained byadding 1 to each of the multiplier, divisor and dividend andsubtracting 1 from the result of calculation at the end tominimize the calculation error, as expressed by equations(107) and (108). However, because thenumber 1 is LSB whichhas no significant effect on the result of calculation,subtraction of 1 at the end may be omitted for the simplificationof the circuit.Embodiment 62 is an example of suchsimplification, where YH' signal at the position GC of rows, column t is calculated by equation (111) below.YH'(s,t) = (GC(s,t)+1) × ((Y'LPF(s,t)+1)/(GCLPF(s,t)+1))
• The YH' signal at the position (s, t) of the colorfilter of kind K of the pixel of interest is given by theequation (112) below.YH'(s,t) = (K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1))
Embodiment 63
• Fig.66 shows a block circuit diagram of the color videocamera inembodiment 63. In Fig.66, symbols which are thesame as those in Fig.63 indicate the identical portions.Numerals 29, 30 represent a lookup table for logarithm and alookup table for power which are similar to those shown inFig.27 (embodiment 10).
• The operation will now be described below. The calculationof the YH' signal in case the kind of the color filterof the pixel at position (s, t) is K is given by equation(112) as described inembodiment 62. Logarithmic conversionwith base x as shown in equation (113) is applied to equation(112), where ^ represents power.YH'(s,t) = X^log x{(K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1))}= X^{log x(K(s,t)+1) + log x(Y'LPF(s,t)+1)- log x(KLPF(s,t)+1)}
• In thisembodiment 63, the calculation can be carriedout with lookup tables of small capacity as in the case ofembodiment 10. Although the above description is for the case of calculating YH' signal based on the equation ofembodiment 62, the calculation by means of a lookup tablefor logarithm and a lookup table for power may be applied toembodiment 53.
Embodiment 64
• Fig.67 shows a block circuit diagram of the color videocamera inembodiment 64. In Fig.67, numerals which are thesame as those in Fig.35 indicate the identical portions. InFig.67, numeral 68 represents a low-pass filter (LPF),numeral 70, 71 represent adders, numeral 73 represents ademultiplexer. and numeral 74 represents a band-pass filter.(BPF).
• The operation will now be described below. The basicoperation from thelens 1 to thearithmetic logic unit 75 isthe same as that ofembodiment 22. The output signals of A/Dconverters 59, 61 are mixed in theadder 71. Thedemultiplexer73 switches the composite signal and the output signal ofA/D converter 60 alternately to supply input to the low-passfilter 68, thereby to obtain YL signal as the output. The YLsignal is similar to the luminance signal obtained with theprior art (see Fig.3).
• The operation of thearithmetic logic unit 75 will bedescribed below. The composition of thearithmetic logicunit 75 is the same as that of embodiment 22 (Fig.36). When the low-pass filter output of Y' signal is fed to thedivider82 as the dividend, the arithmetic operation similar tothat inembodiment 22 is performed to obtain the output ofthemultiplier 83 as the YH' signal.
• In Fig.37, YH' signal at the position of green pixel ofcolumn t, for example, is calculated by equation (114)below.YH'(t) = G(t) × (Y'LPF(t)/GLPF(t))
• The YH' signal at the position t of a pixel of kind Kis calculated by equation (115) below, similarly toembodiment22.YH'(t) = K(t) × (Y'LPF(t)/KLPF(t))
• The aperture correction is performed by taking harmonicscomponent YH from the YH' signal by means of the band-passfilter 74, combining the harmonics component YH and theYL signal in theadder 70 and thereby obtaining the luminancesignal Y.
Embodiment 65
• Fig.68 shows a block circuit diagram of the color videocamera inembodiment 65. In Fig.68, numerals which are thesame as those of Fig.67 or Fig.43 indicate the identicalportions. The composition of thearithmetic logic unit 85 isthe same as that of embodiment 23 (Fig.44).
• In this embodiment, too, YH' signal which is similar tothat ofembodiment 64 is obtained from thearithmetic logic unit 85, with aperture correction being carried out similarlytoembodiment 64 thereafter.
• The principle of calculating the YH' signal inembodiments64, 65 is basically the same as the calculation of Ysignal inembodiments 22, 23. When the YH' signal is calculated,it is made possible to eliminate the modulated componentsof the color signal without reducing the harmonics ofthe luminance signal. Thus it is made possible to carry outaperture correction without unnatural enhancement, by takingharmonics component YH from the YH' signal and mixing itwith the YL signal.
Embodiment 66
• Although the composition of the color video camera inembodiment 66 is the same as that of Fig.67, the signalprocessing operation in thearithmetic logic unit 75 isdifferent. Specifically, YH'(t) at the position of G ofcolumn t is calculated by equation (116).YH'(t) = (G(t)+1) × ((Y'LPF(t)+1)/(GLPF(t)+1))-1
• The calculation of the YH' signal at the position t ofa pixel of kind K is given by equation (117) below.YH'(t) = (K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1)) - 1
Embodiment 67
• Fig.69 shows a block circuit diagram of the color videocamera inembodiment 67. In Fig.69, numerals which are thesame as those in Fig.67 indicate the identical portions, andnumeral 27 represents a comparator similar to that shown inFig.22 (embodiment 4).
• The operation will now be described below. The outputsignals of appropriate pixels in the vicinity of the pixelof interest are supplied frommemories 62, 63 to thecomparator27. In a portion where the image has a high spatialfrequency, YH' signal is calculated similarly toembodiment64 and, in a portion of a low spatial frequency, YH' signalis calculated from the weighted averaging value of the Gsignal and RB signal.
• Thearithmetic logic unit 75 operates similarly toembodiment 64 in the portion of high spatial frequency. Inthe portion of low spatial frequency, YH' signal at theposition of G of column t in Fig.37, and at the positionwhere there is no output signal of column t in Fig.38, forexample, is calculated by equation (118) below.YH'(t) = RB(t-1)/4+G(t)/2+RB(t+1)/4
• The YH' signal, when the pixel of interest is locatedat position t, kind of the pixel of interest is J and thekind of the right and left adjacent pixels of the pixel ofinterest is K, is calculated by equation (119) below.YH'(t) = K(t-1)/4+J(t)/2+K(t+1)/4
Embodiment 68
• Although the composition of the color video camera inembodiment 68 is the same as that of embodiment 67 (Fig.69),the signal processing operations in thearithmetic logicunit 75 and in thecomparator 27 are different.
• Inembodiment 68, the difference between the outputsignals of the right and left adjacent pixels of the pixelof interest is compared to a particular threshold. And it isdetermined that the portion has a high spatial frequency ifthe difference between the output signals is greater thanthe threshold, and the operation ofembodiment 64 is carriedout, and it is determined that the portion has a low spatialfrequency if the difference between the output signals isless than the threshold, and accordingly the operation ofembodiment 67 described above is carried out.
• Results of calculation of the same equations (66) and(67) as inembodiment 26 described previously are comparedto a particular threshold, and the method of calculating theYH' signal is selected according to the result of comparison.
Embodiment 69
• Fig.70 shows a block circuit diagram of the color video camera inembodiment 69. In Fig.70, symbols which are thesame as those in Fig.67 indicate the identical portions, andnumeral 28 represents a lookup table for division similar tothat shown in Fig.23 (embodiment 6).
• The operation of the lookup table fordivision 28 isthe same as that inembodiment 6, and description thereofwill be omitted.
Embodiment 70
• An example in which the method of calculation in embodiment7 with respect toembodiment 6 is applied toembodiment69 described above is thisembodiment 70. The operationof the lookup table fordivision 28 inembodiment 70 is thesame as that in embodiment 7, and description thereof willbe omitted here.
Embodiment 71
• The composition of the color video camera inembodiment71 is the same as that of embodiment 64 (Fig.67). Inembodiment71, a one-dimensional low-pass filter is used as adigital filter which is made up of only bit shift circuitshaving weightings such as 1/8, 1/4. The construction of theone-dimensional low-pass filter is similar to that in embodiment8 (Fig.26) and will not be described here.
Embodiment 72
• Although the composition of the color video camera inembodiment 72 is the same as that of embodiment 64 (Fig.67),the method of calculating the YH' signal in thearithmeticlogic unit 75 is different.
• Inembodiment 66, YH signal is obtained by adding 1 toeach of the multiplier, divisor and dividend and subtracting1 from the result of calculation at the end to minimize thecalculation error, as expressed by equations (116) and(117). However, because thenumber 1 is LSB which has nosignificant effect on the result of calculation, subtractionof 1 at the end may be omitted for the simplification of thecircuit.Embodiment 72 is an example of such simplification,where the YH' signal at the position of G of column t iscalculated by equation (120) below.YH'(t) = (G(t)+1) × ((Y'LPF(t)+1)/(GLPF(t)+1))The calculation of the YH' signal at the position t of apixel of kind K is given by equation (121).YH'(t) = (K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1))
Embodiment 73
• Fig.71 shows a block circuit diagram of the color videocamera inembodiment 73. In Fig.71, numerals which are thesame as those in Fig.67 indicate the identical portions andnumerals 29, 30 represent a lookup table for logarithm and alookup table for power which are similar to those shown inFig.27 (embodiment 10).
• The operation will now be described below. YH' signal in case the kind of the pixel at position t is K, for example,is given by equation (121) as described in embodiment.72. To the equation (121), logarithmic conversion with basex as shown in equation (122) is applied, where ^ representspower.YH'(t) = X^log x {(K(t)+1) × ((Y'LPF(t)+1)/(KLPF(t)+1))}= X^{log x (K(t)+1) + log x(Y'LPF(t)+1)- log x (KLPF(L)+1)}
• Inembodiment 73, too, the calculation can be carriedout by using lookup tables of small capacity as inembodiment10 described previously. Although the above descriptionis for the case of calculating YH' signal based on theequation ofembodiment 72, it goes without saying that thecalculation by means of a lookup table for logarithm and alookup table for power may be applied toembodiment 64.
Embodiment 74
• Fig.72 shows a block circuit diagram of the color videocamera inembodiment 74. In Fig.72, numerals which are thesame as those in Fig.67 indicate the identical portions andwill not be described here. In Fig.72,numerals 92, 93, 94represent two-dimensional memories, andnumerals 95, 96, 97represent two-dimensional low-pass filters (LPF). The compositionof thearithmetic logic unit 75 is the same as that of embodiment 22 (see Fig.36).
• The operation will now be described below. The basicoperation is the same as that ofembodiment 64. Similarly toembodiment 32, the two-dimensional memories 92, 93, 94 storeG signal, RB composite signal and Y' signal written therein(see Fig.49, Fig.50, Fig.51), and two-dimensional low-passfilter outputs (see Fig.52, Fig.53, Fig.54) are obtainedfrom the two-dimensional low-pass filters 95, 96, 97.
• Embodiment 74 is an example of using two-dimensionallow-pass filters instead of one-dimensional low-pass filtersinembodiment 64. In Fig.49, YH' signal at the position ofgreen pixel of row s, column t, for example, is calculatedby equation (123).YH'(s,t) = G(s,t) × (Y'LPF(s,t)/GLPF(s,t))
• The calculation of YH' signal at position (s,t) ofpixel of kind K (K is either G or RB) is given by equation(124) below, similarly toembodiment 32.YH'(s,t) = K(s,t) × (Y'LPF(s,t)/KLPF(s,t))
Embodiment 75
• Fig.73 shows a block circuit diagram of the color videocamera inembodiment 75. In Fig.73, numerals which are thesame as those in Fig.72 indicate the identical portions andwill not be described here. In Fig.73, thearithmetic logicunit 85 has the same composition as that of thearithmeticlogic unit 85 inembodiment 23 described previously (Fig.44). In thisembodiment 75, thedemultiplexer 72 feedsG signal and R, B composite signal to thearithmetic logicunit 85. Inembodiment 75, two-dimensional low-pass filtersare used instead of one-dimensional low-pass filters inembodiment 65.
• The principle of calculating Y signal inembodiments74, 75 is basically the same as that inembodiments 64, 65,and is capable of eliminating the modulated components ofthe color signal without reducing the harmonics of theluminance signal to calculate the YH' signal, thereby enablingit to carry out the aperture correction without unnaturalenhancement by taking the harmonics component YH fromthe YH' signal and mixing it with YL signal.
Embodiment 76
• Although the composition of the color video camera inembodiment 76 is the same as that of Fig.72, the signalprocessing operation in thearithmetic logic unit 75 isdifferent. Specifically, YH'(s,t) at the position of G at(s, t) is calculated by equation (125).YH'(s,t) = (G(s,t)+1) × ((Y'LPF(s,t)+1)/(GLPF(s,t)+1)) - 1
• The calculation of the YH' signal at the position (s,t) of a pixel of kind K is given by equation (126) below.YH'(s,t) = (K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1))-1
Embodiment 77
• Fig.74 shows a block circuit diagram of the color videocamera in embodiment 77. In Fig.74, numerals which are thesame as those in Fig.72 indicate the identical portions, andnumeral 27 represents a comparator similar to that shown inFig.22 (embodiment 4).
• The operation will now be described below. The outputsignals of appropriate pixels in the vicinity of the pixelof interest are supplied from two-dimensional memories 92,93 to thecomparator 27. In a portion where the image has ahigh spatial frequency, YH' signal is calculated similarlytoembodiment 74 and, in a portion of a low spatial frequency,YH' signal is calculated from the weighted averagingvalue of the outputs of N kinds of pixels in the vicinity ofthe pixel of interest.
• Thearithmetic logic unit 75 operates similarly toembodiment 74 in the portion of high spatial frequency. Inthe portion of a low spatial frequency, YH' signal at theposition of G of row s, column t in Fig.49, and at theposition where there is no output signal of row s, column tin Fig.50, for example, is calculated by equation (127)below.YH'(s,t) = RB(s-1,t-1)/16+G(s-1,t)/8+RB(s-1,t+1)/16+RB(s,t-1)/8+G(s,t)/4+RB(s,t+1)/8+RB(s+1,t-1)/16+G(s+1,t)/8+RB(s+1,t+1)/16
• The YH' signal, when the pixel of interest is locatedat position (s, t), kind of the pixel of interest is J andthe kind of the right and left adjacent pixels of the pixelof interest is K, is calculated by equation (128) below.YH'(s,t) = K(s-1,t-1)/16+J(s-1,t)/8+K(s-1,t+1)/16+K(s,t-1)/8+J(s,t)/4+K(s,t+1)/8+K(s+1,t-1)/16+J(s+1,t)/8+K(s+1,t+1)/16
Embodiment 78
• Although the composition of the color video camera inembodiment 78 is the same as that of embodiment 77 (Fig.74),the signal processing operations in thearithmetic logicunit 75 and in thecomparator 27 are different.
• In embodiment 78, the difference between the outputsignals of the right and left adjacent pixels of the pixelof interest or the difference between the output signals ofthe upper and lower adjacent pixels of the pixel of interestis compared to a particular threshold. And it is determinedthat the portion has a high spatial frequency if the differencebetween the output signals is greater than the threshold, and the operation ofembodiment 74 is carried out, and,it is determined that the portion has a low spatial frequencyif the difference between the output signals is less thanthe threshold, and the operation of embodiment 77 describedabove is carried out
• Results of the calculations by the same equations (77)through (80) as in embodiment 36 described previously arecompared to a particular threshold to select one of themethods of generating YH' signal according to the result ofcomparison.
Embodiment 79
• Although the composition of the color video camera inembodiment 79 is the same as that of embodiment 77 (Fig.74),the signal processing operations in thearithmetic logicunit 75 and in thecomparator 27 are different.
• In embodiment 79, the difference between the outputs ofthe pixels of the same kind in the vicinity of the pixel ofinterest is compared to a particular threshold. And it isdetermined that the portion has a high spatial frequency ifthe difference is greater than the threshold, and the operationofembodiment 74 is carried out, and, it is determinedthat the portion has a low spatial frequency if the differenceis less than the threshold, and the operation of embodiment77 described above is carried out.
• Results of the calculations by the same equations (81) through (88) as in embodiment 37 described previously arecompared to a particular threshold to select one of themethods of generating YH' signal according to the result ofcomparison.
Embodiment 80
• Although the composition of the color video camera inembodiment 80 is the same as that of embodiment 77 (Fig.74),the signal processing operations in thearithmetic logicunit 75 and in thecomparator 27 are different.
• Inembodiment 80, the difference between the outputsignals of the diagonally adjacent pixels interposing iscompared to a particular threshold. And it is determinedthat the portion has a high spatial frequency if the differencebetween the output signals is greater than the threshold,and the operation ofembodiment 74 is carried out, and,it is determined that the portion has a low spatial frequencyif the difference between the output signals is less thanthe threshold, and the operation of embodiment 77 describedabove is carried out.
• Results of the calculations by the same equations (89)through (92) as in embodiment 38 described previously arecompared to a particular threshold to select one of themethods of generating YH' signal according to the result ofcomparison.
Embodiment 81
• Fig.75 shows a block circuit diagram of the color videocamera inembodiment 81. In Fig.75, symbols which are thesame as those in Fig.72 indicate the identical portions, andnumeral 28 represents a lookup table for division similar tothat shown in Fig.23 (embodiment 6).
• The operation of the lookup table fordivision 28 isthe same as that inembodiment 6, and description thereofwill be omitted.
Embodiment 82
• An example in which the method of calculation in embodiment7 with respect toembodiment 6 is applied toembodiment81 described above is thisembodiment 82. The operationof the lookup table fordivision 28 inembodiment 82 is thesame as that in embodiment 7, and description thereof willbe omitted.
Embodiment 83
• The composition of the color video camera inembodiment83 is similar to that of embodiment 74 (Fig.72). Inembodiment83, a two-dimensional low-pass filter made up of onlybit shift circuits of weightings such as 1/8, 1/4, 1/2 isused as a digital filter. The composition of the two-dimensionallow-pass filter is similar to that of embodiment 19 (Fig.33) where output Y'LPF(s, t) is obtained from thetwo-dimensional low-pass filter in response to the synthesizedsignal Y', and description thereof will be omitted. Itshould be noted here, however, that use of a two clock delaycircuit instead of a one clock delay circuit serves thepurpose, because the two-dimensional low-pass filter for theKth pixel at the position (s, t) of the pixel of interest ismade by arranging the Kth pixels alternately every otherpixel, in the horizontal direction.
Embodiment 84
• Although the composition of the color video camera inembodiment 84 is the same as that of embodiment 74 (Fig.72),the method of calculating the YH' signal in thearithmeticlogic unit 75 is different.
• In embodiment 76, luminance signal is obtained byadding 1 to each of the multiplier, divisor and dividend andsubtracting 1 from the result of calculation at the end tominimize the calculation error, as expressed by the equations(125) and (126). However, because thenumber 1 is LSBwhich has no significant effect on the result of calculation,subtraction of 1 at the end may be omitted for thesimplification of the circuit. Embodiment 84 is an exampleof such simplification, and YH' signal at the position of Gof row s, column t is calculated by equation (129) below.YH'(s,t) = (G(s,t)+1) × ((Y'LPF(s,t)+1)/(GLPF(s,t)+1))
• YH' signal at position (s,t) of pixel of kind K isgiven by the equation (130) below.YH'(s,t) = (K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1))
Embodiment 85
• Fig.76 shows a block circuit diagram of the color videocamera inembodiment 85. In Fig.76, symbols which are thesame as those in Fig.72 indicate the identical portions.Numerals 29, 30 represent a lookup table for logarithm and alookup table for power similar to those shown in Fig.27(embodiment 10).
• The operation will now be described below. The calculationof the YH' signal in case the kind of the pixel atposition (s,t) is K is given, for example, by the equation(130) as described in embodiment 84. To this equation (130),the logarithmic conversion with base x is applied as expressedby equation (131), where ^ represents power.YH'(s,t) = X^ log x {(K(s,t)+1) × ((Y'LPF(s,t)+1)/(KLPF(s,t)+1))}= X^{log x (K(s,t)+1) + log x (Y'LPF(s,t)+1)- log x (KLPF(s,t)+1)}
• In thisembodiment 85, the arithmetic operation can becarried out by using lookup tables of small capacity similarly toembodiment 10 described before. Although the abovedescription is for the case of calculating YH' signal basedon the equation of embodiment 84, the calculation by meansof a lookup table for logarithm and a lookup table for powermay be applied toembodiment 74.
• Although the image sensor in the above embodiments isof a type which reads two upper and lower adjacent pixels bymixing them, they may be of a type which reads every pixelseparately. The criterion of determining the level of spatialfrequency of an image is also not restricted to theembodiments described above and may be otherwise.
• Although descriptions of the above embodiments assumecomplementary colors for the kind of color filters in theembodiments which use a single image sensor, color filtersof primary colors or combination of a primary color and acomplementary color may be used. Although examples usingfield memories for the color separation memories are given,line memories of the number of lines required for the processingof low-pass filters may be used.
• Although the synthesized signal Y' of G signal and R, Bcomposite signal is synthesized by a demultiplexer in theembodiments by using three types of image sensors, use ofadders or other means may be used to synthesize the signals.
• Although harmonics component is taken from the luminancesignal by means of the band-pass filter in the embodiment where the aperture correction is carried out, the high-pass filter may beused for this purpose. Embodiments where the YL signal is processed as digitalsignal are given, but it may be processed as analog signal.
• As this invention may be embodied in several forms without departingfrom the essential characteristics thereof, the present embodiments are thereforeillustrative and not restrictive, since the scope of the invention is defined by theappended claims rather than by the description preceding them.

Claims (35)

1. A color video camera for obtaining color images comprising:

   an image sensor (2) comprising photoelectric transducers (53, 54, 55)arranged in at least one two-dimensional plane for providing image signalsrepresenting different spectral characteristics;

   low-pass filters (20, 21, 65, 66) for filtering said image signals,wherein the image signal representing a given spectral response characteristicis input to a respective low-pass filter, to obtain 1 st through Nth filter outputs;

   a further low-pass filter (22, 67) for filtering outputs from the imagesensor to produce a low-pass filter luminance signal representative of theluminance;

      characterised in that the camera also comprises:
      first means (23, 75) for calculating a luminance signal at coordinatesof interest by multiplying the image signal for the coordinates of interest,input to the Kth low-pass filter, plus a constant value by the ratio of the low-passfilter luminance signal for the coordinates of interest plus the constantvalue to the Kth low-pass filter output for the coordinates of interest plus theconstant value.
2. A color video camera as claimed in claim 1 comprising:

   a spectral optical system (52) for decomposing an incident ray into Nkinds of light;

   means (56, 57, 58) for amplifying signals obtained from the imagesensors;

   means (72) for synthesizing signals output from the image sensor toobtain a signal representative of the luminance and so to derive said low-passfilter luminance signal;

   and wherein the photoelectric transducers (53, 54, 55) are opticallystaggered relative to each other.
3. A color video camera as claimed in claim 1 or claim 2, wherein
      the first calculating means (23, 75) applies logarithmic conversion tothe fractional component and uses a lookup table for logarithm and a lookuptable for power.
4. A color video camera as claimed in claim 1 or claim 2, wherein
      the constant value is subtracted from the result of the multiplicationand division, thereby to obtain the luminance signal.
5. A color video camera as claimed in claim 1, wherein
      the constant value is zero so that the image signal for the coordinatesof interest input to the Kth (1 ≦ K ≦ N) low-pass filter is multiplied by theratio of the low-pass filter luminance signal for the coordinates of interest tothe Kth low-pass filter output for the coordinates of interest, thereby to obtainthe luminance signal.
6. A color video camera as claimed in claim 5, wherein

   the first means (23) for calculating the luminance signal is for aportion of the image having a high spatial frequency, and the camera furthercomprises:

   second means for calculating the luminance signal, from the weightedaverage value of outputs in the vicinity of the coordinates of interest, in aportion of the image having a low spatial frequency.
7. A color video camera as claimed in claim 5, comprising:

   second means for calculating the luminance signal from the weightedaverage value of outputs in the vicinity of the coordinates of interest; and

   means for selecting the first or the second calculating means,depending on whether the difference between the outputs of right and leftadjacent portions of the coordinates of interest is greater than a predeterminedthreshold or not.
8. A color video camera as claimed in claim 5, wherein
      the fractional product is calculated using a lookup table (28) fordivision, and, assuming that the input to the table is n-bit data, the input ischecked to see if the most significant bit thereof is 1 and, if it is 0, the next bitis checked subsequently eventually stopping at the nth bit, and bit shiftoperation is carried out after this check, thereby using the effective n-bit data with bits 0 removed therefrom as the input to the lookup table (28) fordivision.
9. A color video camera as claimed in claim 8, wherein
      the content of the lookup table (28) for division is increased by n bitsthrough bit shift operation on the result of the dividing operation, to take theintegral part and discarding the fractional part, while applying bit shiftoperation to the result of calculation which uses the value, to reduce it by nbits to return it to the original number of bits, thereby taking the integral partas the result of calculation.
10. A color video camera as claimed in claim 5, wherein
       the calculating means applies logarithmic conversion to the fractionalcomponent and uses a lookup table for the logarithm values and a lookuptable for the power values.
11. A color video camera as claimed in claim 5, wherein
       each of the low-pass filters (20, 21, 65, 66) includes a plurality of bitshift circuits (105, 106, 107, 108, 109) and an adder (110) which takes in theoutputs of the bit shift circuits (105, 106, 107, 108, 109) as the input thereof.
12. A color video camera as claimed in any one of claims 1 to 6 or8 to 11 wherein the low-pass filters are two-dimensional low-pass filters (14,15, 16, 17).
13. A color video camera as claimed in claim 5, wherein the low-passfilters are two-dimensional low-pass filters (14, 15, 16, 17), and

    comprising second calculating means for calculating the luminancesignal from the weighted average value of outputs in the vicinity of thecoordinates of interest; and

    means for selecting the first or the second calculating means,depending on whether either the difference between the outputs of right andleft adjacent portions of the coordinates of interest or the difference betweenthe outputs of upper and lower adjacent portions of the coordinates of interestis greater than a predetermined threshold or not.
14. A color video camera as claimed in claim 5, wherein the low-passfilters are two-dimensional low-pass filters (14, 15, 16, 17), and

    comprising second calculating means for calculating the luminancesignal from the weighted average value of outputs in the vicinity of thecoordinates of interest; and

    means for selecting the first or the second calculating means,depending on whether the difference between the outputs for the same kind of spectral response characteristic in the vicinity of the coordinates of interest isgreater than a predetermined threshold or not.
15. A color video camera as claimed in claim 5, wherein the low-passfilters are two-dimensional low-pass filters (14, 15, 16, 17), and

    comprising second calculating means for calculating the luminancesignal from the weighted average value of outputs in the vicinity of thecoordinates of interest; and

    means for selecting the first or the second calculating means,depending on whether either the difference between the outputs of upper-leftdiagonally adjacent portion and the lower-right diagonally adjacent portion ofthe coordinates of interest or the difference between the outputs of the lower-leftdiagonally adjacent portion and the upper-right diagonally adjacentportion of the coordinates of interest is greater than a predetermined thresholdor not.
16. A color video camera as claimed in any one of claims 1 to 15wherein signals output from transducers having different spectralcharacteristics are mixed to produce said image signals.
17. A color video camera for obtaining color images comprising:

    an image sensor (2) comprising photoelectric transducers for providingimage signals representing different spectral characteristics;

    low-pass filters (20, 21, 65, 66) for filtering said image signals,wherein the image signal representing a given spectral response characteristicis input to a respective low-pass filter, to obtain 1 st through Nth filter outputs;

    a further low-pass filter (22, 67) for filtering outputs from the imagesensor to produce a low-pass filter luminance signal representative of theluminance;

       characterised in that the camera also comprises:

    means (23, 75) for calculating a first signal for coordinates of interestby multiplying the image signal for the coordinates of interest, input to theKth low-pass filter, plus a constant value by the ratio of the low-pass filterluminance signal for the coordinates of interest plus the constant value to theKth low-pass filter output for the coordinates of interest plus the constantvalue;

    a band-pass filter (46, 74) wherein the calculated first signal issupplied as the input to obtain a second signal for aperture correction;

    a still further low-pass filter (45, 68) for filtering outputs from theimage sensor to produce a third signal; and

    means (47, 70) for obtaining a luminance signal by combining thesecond signal and the third signal.
18. A color video camera as claimed in claim 17 comprising:

    a spectral optical system (52) for decomposing an incident ray into Nkinds of light;

    means (56, 57, 58) for amplifying signals obtained from the imagesensor;

    means (72) for synthesizing signals output from the image sensor toobtain a signal representative of the luminance and so to derive said low-passfilter luminance signal;

    and wherein the photoelectric transducers (53, 54, 55) are opticallystaggered relative to each other.
19. A color video camera as claimed in claim 17 or claim 18,wherein
       the calculating means (23, 75) applies logarithmic conversion to thefractional component and uses a lookup table for the logarithm values and alookup table for the power values.
20. A color video camera as claimed in claim 17 or claim 18,wherein the constant value is subtracted from the result of the multiplicationand division, thereby to obtain the first signal.
21. A color video camera as claimed in claim 17 or claim 18,wherein the constant value is zero so that the image signal for the coordinatesof interest input to the Kth (1 ≦ K ≦ N) low-pass filter is multiplied by theratio of the first low-pass filter luminance signal for the coordinates of interest to the Kth first low-pass filter output for the coordinates of interest, thereby toobtain the first signal.
22. A color video camera as claimed in claim 21, wherein

    the first means for calculating the first signal is for a portion of theimage having a high spatial frequency;

    and the camera further comprises second means for calculating thefirst signal, from the weighted average value of outputs in the vicinity of thecoordinates of interest, in a portion of the image having a low spatialfrequency.
23. A color video camera as claimed in claim 21,

    comprising second calculating means for calculating the first signalfrom the weighted average value of the outputs in the vicinity of the pixel ofinterest; and

    means for obtaining the first signal by selecting the first or the secondcalculating means, depending on whether the difference between the outputsof the right and left adjacent portions of the coordinates of interest is greaterthan a predetermined threshold or not.
24. A color video camera as claimed in claim 21, wherein
       the fractional product is calculated using a lookup table (28) fordivision, and, assuming that the input to the table is n-bit data, the input is checked to see if the most significant bit thereof is 1 and, if it is 0, the next bitis checked subsequently eventually stopping at the nth bit, and bit shiftoperation is carried out after this check, thereby using the effective n-bit datawith bits 0 removed therefrom as the input to the lookup table for division.
25. A color video camera as claimed in claim 24, wherein
       the content of the lookup table for division is increased by n bitsthrough bit shift operation on the result of the dividing operation, to take theintegral part and discarding the fractional part, while applying bit shiftoperation to the result of calculation which uses the value, to reduce it by nbits to return it to the original number of bits, thereby taking the integral partas the result of calculation.
26. A color video camera as claimed in claim 21, wherein
       the calculating means (23, 75) applies logarithmic conversion to thefractional component and uses a lookup table for the logarithm values and alookup table for the power values.
27. A color video camera as claimed in claim 21, wherein
       each of the first low-pass filters (20, 21, 65, 66) includes a plurality ofbit shift circuits (105, 106, 107, 108, 109) and an adder (110) which takes inthe outputs of the bit shift circuits (105, 106, 107, 108, 109) as the inputthereof.
28. A color video camera as claimed in any one of claims 17 to 22or 24 to 27 wherein the first low-pass filters are two-dimensional filters (14,15, 16, 17).
29. A color video camera as claimed in claim 21 wherein the firstlow-pass filters are two-dimensional low-pass filters and comprising:

    second calculating means for calculating the first signal from theweighted average value of outputs in the vicinity of the coordinates ofinterest; and

    means for obtaining the first signal by selecting the first or the secondcalculating means, depending on whether either the difference between theoutputs of the right and left adjacent portions of the coordinates of interest orthe difference between outputs of upper and lower adjacent portions of thecoordinates of interest is greater than a predetermined threshold or not.
30. A color video camera as claimed in claim 21, wherein the firstlow-pass filters are two-dimensional low-pass filters and comprising:

    second calculating means for calculating the first signal from theweighted average value of outputs in the vicinity of the coordinates ofinterest; and

    means for obtaining the first signal by selecting the first or the secondcalculating means, depending on whether the difference between outputs for the same kind of spectral response characteristic in the vicinity of thecoordinates of interest is greater than a predetermined threshold or not.
31. A color video camera as claimed in claim 21, wherein the firstlow-pass filters are two-dimensional low-pass filters and comprising:

    second calculating means for calculating the first signal from theweighted average value of outputs in the vicinity of the coordinates ofinterest; and

    means for obtaining the first signal by selecting the first or the secondcalculating means, depending on whether either the difference betweenoutputs of the upper-left diagonally adjacent portion and the lower-rightdiagonally adjacent portion of the coordinates of interest or the differencebetween the outputs of the lower-left diagonally adjacent portion and theupper-right diagonally adjacent portion of the coordinates of interest is greaterthan a predetermined threshold or not.
32. A color video camera as claimed in any one of claims 17 to 31wherein signals output from transducers having different spectralcharacteristics are mixed to produce said image signals.
33. A color video camera for obtaining color images comprising:

    an image sensor (2) comprising photoelectric transducers for providingimage signals representing different spectral characteristics;

    two-dimensional low-pass filters (14, 15, 16, 17) for filtering saidimage signals, wherein the image signal representing a given spectral responsecharacteristic is input to a respective low-pass filter, to obtain 1 st through Nthfilter outputs;

       characterised in that the camera also comprises:
       means (42) for calculating a luminance signal for coordinates ofinterest, wherein the image signal for the coordinates of interest, input to theJth low-pass filter, is multiplied by the ratio of the Kth low-pass filter outputfor the coordinates of interest to the Jth two-dimensional low-pass filteroutput for the coordinates of interest, to calculate the contribution to theluminance signal associated with the Kth spectral response characteristic forthe coordinates of interest, and the contributions for the other spectralresponse characteristics are calculated similarly for the coordinates of interest,and used to calculate the luminance signal for the coordinates of interest.
34. A method of deriving a signal representing luminance of animage from color signals each representing the intensity of a respective imagecolor component in a respective set of regions, the method comprising usinglow-pass filtering to determine the intensity contribution made by each saidcolor relative to the other colors in an area of the image, and calculating aluminance value for each region in accordance with the intensity of the colorsignal for that region and the determined contribution made by the colorassociated with the respective region.
35. A color video camera comprising means (42) for deriving asignal representing luminance of an image from color signals eachrepresenting the intensity of a respective image color component in arespective set of regions, using low-pass filtering to determine the intensitycontribution made by each said color relative to the other colors in an area ofthe image, and calculating a luminance value for each region in accordancewith the intensity of the color signal for that region and the determinedcontribution made by the color associated with the respective region.

Images